@InProceedings{	  aarts93a,
  author	= {E. H. L. Aarts and H. P. Stehouwer},
  title		= {Neural Networks and the Travelling Salesman Problem},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {950--955},
  publisher	= {Springer},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@Article{	  abdallah95a,
  author	= {M. A. Abdallah and T. I. Samu and W. A. Grissom},
  title		= {Automatic target identification using neural networks},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1995},
  volume	= {2588},
  pages		= {556--65},
  note		= {(Intelligent Robots and Computer Vision XIV: Algorithms,
		  Techniques, Active Vision, and Materials Handling Conf.
		  Date: 23--26 Oct. 1995 Conf. Loc: Philadelphia, PA, USA
		  Conf. Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  abdel-aty-zohdy95a,
  author	= {Abdel-Aty-Zohdy, H. S. and Zondy, M. A. },
  title		= {Neural networks for pattern discovery and optimization in
		  signal processing and applications},
  booktitle	= {1995 Canadian Conference on Electrical and Computer
		  Engineering},
  year		= {1995},
  editor	= {Gagnon, F. },
  volume	= {1},
  pages		= {202--6},
  organization	= {Dept. of Electr. \& Syst. Eng. , Oakland Univ. ,
		  Rochester, MI, USA},
  publisher	= {IEE},
  address	= {New York, NY, USA},
  abstract	= {This work is intended to advance both conceptual and
		  implementation techniques of supervised and
		  discovery-driven neural networks in noisy time varying
		  signal processing applications. Successful neural networks
		  and their significance in applications are based on;
		  selection of proper theoretical algorithms for learning,
		  appropriate selection of the sequencing of signal
		  processing tasks, and efficient VLSI system implementation.
		  In this paper, we present pattern discovery Self Organizing
		  Feature Map (SOFM), followed by a Recurrent Dynamic Neural
		  Network (RDNN) algorithm for signal representation and
		  processing. This approach combines the benefits of RDNNs
		  with SOFM counter part. Preliminary designs,
		  implementations, test results and validation of
		  silicon-chips for each of the above neural network
		  approaches are also presented.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  abdel-aty-zohdy96a,
  author	= {H. S. Abdel-Aty-Zohdy and M. A. Zohdy},
  title		= {Analog/digital implementation of neural networks for
		  pattern discovery and optimization in signal processing
		  applications},
  booktitle	= {38th Midwest Symposium on Circuits and Systems.
		  Proceedings},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {1},
  editor	= {L. P. Caloba and P. S. R. Diniz and A. C. M. {de Querioz}
		  and E. H. Watanabe},
  address	= {New York, NY, USA},
  pages		= {277},
  dbinsdate	= {oldtimer}
}

@InCollection{	  abdelazim96a,
  author	= {H. Y. Abdelazim},
  title		= {A hybrid fuzzy-neural approach to the recognition of
		  Arabic script},
  booktitle	= {Proceedings of the 5th International Conference and
		  Exhibition on Multi-Lingual Computing},
  publisher	= {Univ. Cambridge},
  year		= {1996},
  address	= {Cambridge, UK},
  pages		= {2/3/1--14},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  abdulkader00a,
  author	= {Abdulkader, Hasan and Ibnkahla, Mohamed},
  title		= {Convergence properties of self-organizing maps applied for
		  communication channel equalization},
  booktitle	= {ICASSP, IEEE International Conference on Acoustics, Speech
		  and Signal Processing---Proceedings},
  year		= {2000},
  editor	= {},
  volume	= {6},
  pages		= {3502--3505},
  organization	= {Natl Polytechnics Inst of Toulouse},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {This paper studies the convergence properties of a Self
		  Organizing Map (SOM) equalizer. The transmitted signal is
		  assumed to be a 4-QAM signal corrupted by an additive white
		  gaussian noise (AWGN). Several theoretical results such as
		  map ordering and stability conditions are given in the
		  paper. These results show good fit with computer
		  simulations.},
  dbinsdate	= {2002/1}
}

@Article{	  abel96a,
  author	= {Abel, E. W. and Zacharia, P. C. and Forster, A. and
		  Farrow, T. L. },
  title		= {Neural network analysis of the {EMG} interference
		  pattern},
  journal	= {Medical Engineering \& Physics},
  year		= {1996},
  volume	= {18},
  number	= {1},
  pages		= {12--17},
  dbinsdate	= {oldtimer}
}

@Article{	  abidi94a,
  author	= {Abidi, M. A. and Yasuki, S. and Crilly, P. B. },
  title		= {Image compression using hybrid neural networks combining
		  the auto-associative multi-layer perceptron and the
		  \mbox{self-organizing} feature map},
  journal	= {IEEE Transactions on Consumer Electronics},
  year		= {1994},
  volume	= {40},
  number	= {4},
  pages		= {796--811},
  month		= {Nov},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@Article{	  abidi94b,
  author	= {Abidi, M. A. and Yasuki, S. and Crilly, P. B.},
  title		= {Image compression using hybrid neural networks},
  journal	= {Digest of Technical Papers---IEEE International Conference
		  on Consumer Electronics},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1994},
  number	= {},
  volume	= {},
  pages		= {70--71},
  abstract	= {A new image compression technique using hybrid neural
		  networks combining two different learning network-s (1)
		  AMLP (Auto-associative multi-layer perceptron) and (2) SOFM
		  (self-organizing feature map) is presented. The neural
		  networks simultaneously perform dimensionality reduction
		  with AMLP and categorization with SOFM to compress the
		  image. The two hybrid neural networks which form a parallel
		  and a serial architecture are examined using theoretical
		  analysis and computer simulation. These hybrid neural
		  networks achieve clear performance improvement with respect
		  to decoded picture quality and compression ratios, compared
		  to existing image compression techniques.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  abidi96a,
  author	= {S. S. R. Abidi},
  title		= {Neural networks and child language development: a
		  simulation using a modular neural network architecture},
  booktitle	= {ICNN 96. The 1996 IEEE International Conference on Neural
		  Networks},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {2},
  address	= {New York, NY, USA},
  pages		= {840--5},
  dbinsdate	= {oldtimer}
}

@Article{	  abidi97a,
  author	= {S. S. R. Abidi},
  title		= {Using neural networks to explicate human category
		  learning: a simulation of concept learning and
		  lexicalisation},
  journal	= {Malaysian Journal of Computer Science},
  year		= {1997},
  volume	= {10},
  number	= {2},
  pages		= {60--71},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  aboelela94a,
  author	= {Aboelela, M. A. S. },
  title		= {Short term forecasting of electric daily loads},
  booktitle	= {MEPCON 94. Middle East Power System Conference.
		  Proceedings},
  year		= {1994},
  pages		= {13--17},
  organization	= {Fac. of Eng. , Cairo Univ. , Giza, Egypt},
  publisher	= {Cairo Univ},
  address	= {Giza, Egypt},
  dbinsdate	= {oldtimer}
}

@Article{	  abou01a,
  author	= {Abou Chadi, F. E. Z. and Nashar, A. and Saad, M.},
  title		= {Automatic analysis and classification of surface
		  electromyography},
  journal	= {Frontiers-of-Medical-and-Biological-Engineering},
  year		= {2001},
  volume	= {11},
  pages		= {13--29},
  abstract	= {Parametric modeling of surface electromyography (EMG)
		  algorithms that facilitates automatic SEMG feature
		  extraction and artificial neural networks (ANN) are
		  combined for providing an integrated system for the
		  automatic analysis and diagnosis of myopathic disorders.
		  Three paradigms of ANN were investigated: the multilayer
		  backpropagation algorithm, the self-organizing feature map
		  algorithm and a probabilistic neural network model. The
		  performance of the three classifiers was compared with that
		  of the old Fisher linear discriminant (FLD) classifiers.
		  The results have shown that the three ANN models give
		  higher performance. The percentage of correct
		  classification reaches 90%. Poorer diagnostic performance
		  was obtained from the FLD classifier. The system presented
		  here indicates that surface EMG, when properly processed,
		  can be used to provide the physician with a diagnostic
		  assist device.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  abramson01a,
  author	= {Abramson, M. and Wechsler, H.},
  title		= {Competitive reinforcement learning for combinatorial
		  problems},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2001},
  editor	= {},
  volume	= {4},
  pages		= {2333--2338},
  organization	= {George Mason University},
  publisher	= {},
  address	= {},
  abstract	= {This paper shows that the competitive learning rule found
		  in Learning Vector Quantization (LVQ) serves as a promising
		  function approximator to enable reinforcement learning
		  methods to cope with a large decision search space, defined
		  in terms of different classes of input patterns, like those
		  found in the game of Go. In particular, this paper
		  describes S[arsa]LVQ, a novel reinforcement learning
		  algorithm and shows its feasibility for Go. As the
		  distributed LVQ representation corresponds to a (quantized)
		  codebook of compressed and generalized pattern templates,
		  the state space requirements for online reinforcement
		  methods are significantly reduced, thus decreasing the
		  complexity of the decision space and consequently improving
		  the play performance. As a result of competitive learning,
		  SLVQ can win against heuristic players and starts to level
		  off against stronger opponents such as Wally. SLVQ
		  outperforms S[arsa]Linear when playing against both a
		  heuristic player and Wally. Furthermore, while playing Go,
		  SLVQ learns to stay alive while SLinear fails to do so.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  abrantes93a,
  author	= {Abrantes, A. J. and Marques, J. S. },
  title		= {A common framework for snakes and {K}ohonen networks},
  booktitle	= {Neural Networks for Signal Processing III, Proceedings of
		  the 1993 IEEE-SP Workshop},
  year		= {1993},
  editor	= {Kamm, C. A. and Kuhn, G. M. and Yoon, B. and Chellappa, R.
		  and Kung, S. Y. },
  pages		= {251--60},
  organization	= {INESC/ISEL, Lisbon, Portugal},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  abrantes95a,
  author	= {Abrantes, A. J. and Marques, J. S. },
  title		= {Unified approach to snakes, elastic nets and {K}ohonen
		  maps},
  booktitle	= {Proceedings of the 1995 International Conference on
		  Acoustics, Speech, and Signal Processing},
  year		= {1995},
  volume	= {5},
  pages		= {3427--30},
  organization	= {INESC, Polytech. Inst. of Lisbon, Portugal},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  abrantes95b,
  author	= {A. J. Abrantes and J. S. Marques},
  title		= {Exploiting the common structure of {SOM} edge linking
		  algorithms: an experimental study},
  booktitle	= {Proceedings of the International Conference on Image
		  Processing},
  publisher	= {IEEE Computer Society Press},
  year		= {1995},
  volume	= {3},
  address	= {Los Alamitos, CA, USA},
  pages		= {624--7},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  abuelgasim94a,
  author	= {Abuelgasim, A. and Gopal, S. },
  title		= {Classification of multiangle and multispectral {ASAS} data
		  using a hybrid neural network model},
  booktitle	= {IGARSS '94. International Geoscience and Remote Sensing
		  Symposium. Surface and Atmospheric Remote Sensing:
		  Technologies, Data Analysis and Interpretation},
  year		= {1994},
  volume	= {3},
  pages		= {1670--2},
  organization	= {Dept. of Geogr. , Boston Univ. , MA, USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  acciani93a,
  author	= {Acciani, G. and Bellomo, A. and Chiarantoni, E. and
		  Paradiso, A. },
  title		= {Validation of neural network analysis to predict prognosis
		  in breast cancer patients},
  booktitle	= {Proceedings of the 36th Midwest Symposium on Circuits and
		  Systems},
  year		= {1993},
  volume	= {1},
  pages		= {453--6},
  organization	= {Dipartimento di Elettrotecnica ed Elettronica, Politecnico
		  di Bari, Italy},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  acciani96a,
  author	= {G. Acciani and E. Chiarantoni and M. Minenna and F.
		  Vacca},
  title		= {Multivariate Data Projection Techniques based on a Network
		  of Enhanced Neural Elements},
  booktitle	= {ICNN'96. The 1996 IEEE International Conference on Neural
		  Networks},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {1},
  address	= {New York, NY, USA},
  pages		= {211--216},
  dbinsdate	= {oldtimer}
}

@Article{	  acharya97a,
  author	= {S. Acharya and R. Sadananda},
  title		= {Promoting software reuse using \mbox{self-organizing}
		  maps},
  journal	= {Neural Processing Letters},
  year		= {1997},
  volume	= {5},
  number	= {3},
  pages		= {219--26},
  dbinsdate	= {oldtimer}
}

@Article{	  addison88a,
  author	= {Edwin R. Addison and William Dedmond},
  title		= {Criteria for choosing connectionist paradigms for
		  real-time data fusion and adaptive discrimination},
  journal	= {Neural Networks},
  year		= {1988},
  volume	= {1},
  number	= {1 SUPPL},
  pages		= {419},
  dbinsdate	= {oldtimer}
}

@Article{	  ae91a,
  author	= {T. Ae},
  title		= {Neural networks and functional memories},
  journal	= {Joho Shori},
  year		= {1991},
  volume	= {32},
  number	= {12},
  pages		= {1301--1309},
  note		= {(in Japanese)},
  x		= {Discusses artificial neural networks; . . . Kohonen
		  networks; learning vector quantization; topology-preserving
		  maps;},
  dbinsdate	= {oldtimer}
}

@Article{	  ae93a,
  author	= {Tadashi Ae and Reiji Aibara},
  title		= {Non von {N}eumann Chip Architecture---Present and Future},
  journal	= {IEICE Trans. Elecronics},
  year		= {1993},
  volume	= {E76-C},
  number	= {7},
  pages		= {1034--1044},
  dbinsdate	= {oldtimer}
}

@InCollection{	  ae96a,
  author	= {T. Ae and K. Sakai and T. Toyosaki},
  title		= {Neural artificial intelligence system},
  booktitle	= {14th International Congress on Cybernetics. Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1996},
  editor	= {J. Parisi and S. C. Muller and W. Zimmermann},
  address	= {Berlin, Germany},
  pages		= {471--6},
  dbinsdate	= {oldtimer}
}

@InCollection{	  aggarwal96a,
  author	= {R. K. Aggarwal and Q. Y. Xuan and A. T. Johns},
  title		= {Fault classification for double-circuits using
		  self-organization mapping},
  booktitle	= {32nd Universities Power Engineering Conference. UPEC '97},
  publisher	= {ACM},
  year		= {1996},
  volume	= {1},
  editor	= {E. A. Fox and G. Marchionini},
  address	= {New York, NY, USA},
  pages		= {440--3},
  dbinsdate	= {oldtimer}
}

@Article{	  aghajan93a,
  author	= {H. K. Aghajan and C. D. Schaper and T. Kailath},
  title		= {Machine vision techniques for subpixel estimation of
		  critical dimensions},
  journal	= {Opt. Eng. },
  year		= {1993},
  volume	= {32},
  number	= {4},
  pages		= {828--839},
  month		= {April},
  annote	= {{SOFM} is investigated for edge detection purposes. },
  dbinsdate	= {oldtimer}
}

@Article{	  aguilera01a,
  author	= {Aguilera, P. A. and Frenich, A. G. and Torres, J. A. and
		  Castro, H. and Vidal, J. L. M. and Canton, M},
  title		= {Application of the Kohonen neural network in coastal water
		  management: Methodological development for the assessment
		  and prediction of water quality},
  journal	= {WATER RESEARCH},
  year		= {2001},
  volume	= {35},
  number	= {17},
  month		= {DEC},
  pages		= {4053--4062},
  abstract	= {Kohonen neural network (KNN) was applied to nutrient data
		  (ammonia, nitrite, nitrate and phosphate) taken from
		  coastal waters in a Spanish tourist area. The activation
		  maps obtained were not sufficient to evaluate and predict
		  the trophic status of coastal waters. To achieve this aim,
		  a new methodology is proposed which uses as its starting
		  point the activation maps obtained from KNN. Firstly, to
		  evaluate the trophic status of the coastal waters. it
		  consists of the development of a quadrat system which
		  enables a better classification than the traditional
		  classification based simply on standardized data. The new
		  classification allows clear differentiation of water
		  quality within the mesotrophic band. Secondly, and in order
		  to use the activation maps as predictive tools, the trophic
		  classification. obtained from activation maps, was
		  transposed onto new activation maps. To do this, the
		  activation maps of the sampling points which defined each
		  trophic group were superimposed. To avoid unnecessary
		  complexity and to facilitate the process, this
		  superimposition was undertaken only where the frequency
		  exceeded 0.05. In this way, four frequency maps related to
		  the trophic status of coastal waters (potentially
		  eutrophic, high mesotrophic, low mesotrophic and
		  oligotrophic) were obtained. There was no loss of relevant
		  information in the new maps thus obtained. These frequency
		  maps served as the basis for the successful prediction of
		  the trophic status of random samples of coastal waters.
		  This methodology, based on KNN, is proposed as a tool to
		  aid the decision-making in coastal water quality
		  management. },
  dbinsdate	= {2002/1}
}

@InProceedings{	  ahalt89a,
  author	= {S. C. Ahalt and T. P. Jung and A. K. Krishnamurthy},
  title		= {Radar target identification using the learning vector
		  quantization neural network},
  booktitle	= {Proc. IJCNN'89, International Joint Conference on Neural
		  Networks},
  year		= {1989},
  volume	= {2},
  pages		= {605},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ahalt90a,
  author	= {S. C. Ahalt and T. Jung and A. K. Krishnamurthy},
  title		= {A comparison of radar signal classifiers},
  booktitle	= {Proc. IEEE International Conference on Systems
		  Engineering},
  year		= {1990},
  pages		= {609--612},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  ahalt90b,
  author	= {Stanley C. Ahalt and Ashok K. Krishnamurty and Prakoon
		  Chen and Douglas E. Melton},
  title		= {Competitive Learning Algorithms for Vector Quantization},
  journal	= {Neural Networks},
  year		= {1990},
  volume	= {3},
  number	= {3},
  pages		= {277--290},
  dbinsdate	= {oldtimer}
}

@Article{	  ahmad01a,
  author	= {Ahmad, K. and Vrusias, B. L. and Ledford, A.},
  title		= {Choosing feature sets for training and testing
		  self-organising maps: A case study},
  journal	= {NEURAL COMPUTING \& APPLICATIONS},
  year		= {2001},
  volume	= {10},
  number	= {1},
  pages		= {56--66},
  abstract	= {Statistical pattern recognition techniques, supervised and
		  unsupervised classification techniques being two good
		  examples here, rely on the computations of similarity and
		  distance metrics. The distances are computed in a
		  multi-dimensional space. The axes of this space in
		  principle relate to the features inherent in the input
		  data. Usually, such features are chosen by neural network
		  developers, thereby introducing a possible bias. A method
		  of automatically generating feature sets is discussed, with
		  specific reference to the categorisation of streams of
		  free-text news items. The feature sets were generated by a
		  procedure that automatically selects a group of keywords
		  based on a lexico-semantic analysis. Three different types
		  of text streams---headlines only, news summaries and full
		  news items including the body of the text---have been
		  categorised using Self-Organising Feature Maps (SOFM). A
		  method for assessing the discrimination ability of a SOFM,
		  based on Fisher's Linear Discriminant Rule suggests that
		  the maps trained on vectors related to summaries only
		  provides a fairly accurate cluster when compared with
		  vectors related to full text. The use of summaries as
		  document surrogates for document categorisation is
		  suggested.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  ahmad99a,
  author	= {Ahmad, K. and Bale, T. A. and Burford, D.},
  title		= {Text classification and minimal-bias training vectors},
  booktitle	= {IJCNN'99. International Joint Conference on Neural
		  Networks. Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {4},
  pages		= {2816--19},
  abstract	= {The categorisation of text using neural networks has been
		  pursued in the context of emerging digital libraries.
		  Frequency of key terms in a document set are used to create
		  training vectors and a feature map can be trained such that
		  texts are topologically ordered. The choice of training
		  vectors and training regimes remains an open question in
		  neural network research. In order to minimise training bias
		  we have developed a method which uses the linguistic
		  characteristics of domain-specific texts for the creation
		  of training vectors. This method has been evaluated by
		  classifying a standard free-text document set (TIPSTER's
		  SUMMAC AP news wire collection) using a Kohonen feature
		  map. This method is particularly relevant to the domain of
		  financial prediction, considering the large volume of news
		  reports available to financial analysts.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ahmed00a,
  author	= {Ahmed, Mohamed and Cooper, Brian and Love, Shaun},
  title		= {Document image segmentation using a two-stage neural
		  network},
  booktitle	= {Proceedings of SPIE---The International Society for
		  Optical Engineering},
  year		= {2000},
  editor	= {},
  volume	= {3962},
  pages		= {25--33},
  organization	= {Lexmark Int Inc},
  publisher	= {SPIE},
  address	= {Bellingham, WA},
  abstract	= {In this paper, we present a new system to segment and
		  label document images into text, halftone images, and
		  background using feature extraction and unsupervised
		  clustering. Each pixel is assigned a feature pattern
		  consisting of a scaled family of differential geometrical
		  invariant features and texture features extracted from the
		  cooccurence matrix. The invariant feature pattern is then
		  assigned to a specific region using a two-stage neural
		  network system. The first stage is a self-organizing
		  principal components analysis (SOPCA) network that is used
		  to project the feature vector onto its leading principal
		  axes found by using principal components analysis (PCA).
		  Using the SOPCA algorithm, we can train the SOPCA network
		  to project our feature vector orthogonally onto the
		  subspace spanned by the eigenvectors belonging to the
		  largest eigenvalues. By doing that we ensure that the
		  vector is represented by a reduced number of effective
		  features. The next step is to cluster the output of the
		  SOPCA network into different regions. This is accomplished
		  using a self-organizing feature-map (SOFM) network. In this
		  paper, we demonstrate the power of the SOPCA-SOFM approach
		  to segment document images into text, halftone, and
		  background.},
  dbinsdate	= {2002/1}
}

@InCollection{	  ahmed97a,
  author	= {Mohamed N. Ahmed and Aly A. Farag},
  title		= {Two-Stage Neural Network for Volume Segmentation of
		  Medical Images},
  booktitle	= {Proceedings of ICNN'97, International Conference on Neural
		  Networks},
  publisher	= {IEEE Service Center},
  year		= 1997,
  address	= {Piscataway, NJ},
  volume	= {III},
  pages		= {1373--1378},
  dbinsdate	= {oldtimer}
}

@InCollection{	  ahmed97b,
  author	= {M. N. {A}hmed and A. A. Farag},
  title		= {{3D} segmentation and labeling using
		  \mbox{self-organizing} {K}ohonen network for volumetric
		  measurements on brain {CT} imaging to quantify {TBI}
		  recovery},
  booktitle	= {Proceedings of the 18th Annual International Conference of
		  the IEEE Engineering in Medicine and Biology Society.
		  `Bridging Disciplines for Biomedicine'},
  publisher	= {IEEE},
  year		= {1997},
  volume	= {2},
  editor	= {H. Boom and C. Robinson and W. Rutten and M. Neuman and H.
		  Wijkstra},
  address	= {New York, NY, USA},
  pages		= {738--9},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ahola99a,
  author	= {Ahola, J. and Alhoniemi, E. and Simula, O.},
  title		= {Monitoring industrial processes using the
		  \mbox{self-organizing} map},
  booktitle	= {SMCia/99 Proceedings of the 1999 IEEE Midnight---Sun
		  Workshop on Soft Computing Methods in Industrial
		  Applications},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {},
  pages		= {22--7},
  abstract	= {In this paper, three process monitoring methods based on
		  the self-organizing map (SOM) are presented: trajectory
		  display, fuzzy response and probabilistic response. These
		  approaches are compared with each other and also
		  demonstrated in two case studies: continuous pulping and
		  hot rolling of steel strips.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ahrns95a,
  author	= {Ingo Ahrns and J{\"{o}}rg Bruske and Gerald Sommer},
  title		= {On-Line Learning with Dynamic Cell Structure},
  booktitle	= {Proc. ICANN'95, International Conference on Artificial
		  Neural Networks},
  year		= {1995},
  editor	= {F. Fogelman-Souli{\'{e}} and P. Gallinari},
  volume	= {II},
  pages		= {141--146},
  publisher	= {EC2},
  address	= {Nanterre, France},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  aiello00a,
  author	= {Aiello, A. and Grimaldi, D. and Daponte, P.},
  title		= {Non-intrusive estimation of the carrier frequency in
		  {GMSK} signals},
  booktitle	= {AUTOTESTCON (Proceedings)},
  year		= {2000},
  editor	= {},
  volume	= {},
  pages		= {621--625},
  organization	= {Universita della Calabria},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {The paper presents an efficient technique for evaluating
		  the carrier frequency in GMSK communication systems. This
		  technique operates in a non-intrusive way and is based on
		  measurement of the frequency shift from the expected value.
		  It utilizes the Learning Vector Quantisation neural network
		  based demodulator for reconstructing the transmitted phase.
		  From this and the received phase the mean value of the
		  frequency shift is estimated. The technique does not
		  require a high frequency sampling rate because the
		  base-band signal is processed. Tests performed on
		  experimental GSMK signals show that the technique is quite
		  attractive and is also fast and more accurate compared to
		  others.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  aiello01a,
  author	= {Aiello, A. and Grimaldi, D. and Veltri, S.},
  title		= {Frequency error measurement in {GMSK} signals in a
		  multipath propagation environment},
  booktitle	= {Conference Record---IEEE Instrumentation and Measurement
		  Technology Conference},
  year		= {2001},
  editor	= {},
  volume	= {1},
  pages		= {507--512},
  organization	= {Dip. di Elettronica, Informatica e Sistemistica,
		  Universita della Calabria},
  publisher	= {},
  address	= {},
  abstract	= {The paper presents an efficient method for evaluating the
		  carrier frequency in GMSK communication systems. This
		  method operates in a non-intrusive way. It utilizes the
		  Learning Vector Quantisation neural network based
		  demodulator for reconstructing the transmitted phases. From
		  these and the expected phases is estimated the carrier
		  frequency error. The method is able to operate both in
		  static and multi-path propagation cases and it does not
		  require a high frequency sampling rate because the
		  base-band signal is processed. In order to apply the method
		  two procedures, PSP (Procedure for Static Propagation) and
		  PMP (Procedure for Multi-path Propagation), are set-up.
		  Tests performed on GMSK signals show that the method is
		  quite attractive, fast and more accurate if compared with
		  other approaches.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  aiello01c,
  author	= {Aiello, A. and Grimaldi, D. and Rapuano, S.},
  title		= {{GMSK} neural network based demodulator},
  booktitle	= {Proceedings of the International Workshop on Intelligent
		  Data Acquisition and Advanced Computing Systems: Technology
		  and Applications. IDAACS'2001. IEEE, Piscataway, NJ, USA},
  year		= {2001},
  volume	= {},
  pages		= {2--6},
  abstract	= {The pattern recognition characteristics of the Artificial
		  Neural Networks are used to realise a real demodulator for
		  Gaussian Minimum Shift-Keying signals, used in the GSM
		  telecommunications. The demodulator utilizes the learning
		  vector quantization (LVQ) neural network. It offers both
		  greater efficiency in demodulating and less sensitivity to
		  noise. In order to solve the problem regarding input signal
		  synchronization, a pre-processing phase is organised. The
		  demodulator prototype has been realised by implementing the
		  pre-processing phase and the LVQ neural network on
		  TMS320C30 digital signal processor. The demodulator has
		  been tested according to the European Telecommunication
		  Standard Institute recommendations.},
  dbinsdate	= {2002/1}
}

@Article{	  aigno98a,
  author	= {Song Aigno and Lu Jiren},
  title		= {Evolutionary programming based {K}ohonen network for
		  passive sonar targets clustering analysis},
  journal	= {Acta Electronica Sinica},
  year		= {1998},
  volume	= {26},
  number	= {7},
  pages		= {128--32},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  aike95a,
  author	= {Guo Aike and Sun Haijian and Yang Xian Yi},
  title		= {A multilayer neural network model for the perception of
		  rotational motion},
  booktitle	= {Proceedings of International Conference on Neural
		  Information Processing (ICONIP `95)},
  year		= {1995},
  editor	= {Zhong, Y. and Yang, Y. and Wang, M. },
  volume	= {1},
  pages		= {121--4},
  organization	= {Inst. of Biophys. , Acad. Sinica, Beijing, China},
  publisher	= {Publishing House of Electron. Ind},
  address	= {Beijing, China},
  dbinsdate	= {oldtimer}
}

@Article{	  aires-de-sousa02a,
  author	= {Aires-de-Sousa, Joao and Gasteiger, Johann},
  title		= {Prediction of enantiomeric selectivity in chromatography:
		  Application of conformation-dependent and
		  conformation-independent descriptors of molecular
		  chirality},
  journal	= {Journal of Molecular Graphics and Modelling},
  year		= {2002},
  volume	= {20},
  number	= {5},
  month		= {},
  pages		= {373--388},
  organization	= {Departamento de Quimica, Campus Fac. de Ciencias e
		  Tecnologia, Universidade Nova de Lisboa},
  publisher	= {},
  address	= {},
  abstract	= {In order to process molecular chirality by computational
		  methods and to obtain predictions for properties that are
		  influenced by chirality, a fixed-length
		  conformation-dependent chirality code is introduced. The
		  code consists of a set of molecular descriptors
		  representing the chirality of a 3D molecular structure. It
		  includes information about molecular geometry and atomic
		  properties, and can distinguish between enantiomers, even
		  if chirality does not result from chiral centers. The new
		  molecular transform was applied to two datasets of chiral
		  compounds, each of them containing pairs of enantiomers
		  that had been separated by chiral chromatography. The
		  elution order within each pair of isomers was predicted by
		  means of Kohonen neural networks (NN) using the chirality
		  codes as input. A previously described
		  conformation-independent chirality code was also applied
		  and the results were compared. In both applications
		  clustering of the two classes of enantiomers (first eluted
		  and last eluted enantiomers) could be successfully achieved
		  by NN and accurate predictions could be obtained for
		  independent test sets. The chirality code described here
		  has a potential for a broad range of applications from
		  stereoselective reactions to analytical chemistry and to
		  the study of biological activity of chiral compounds. },
  dbinsdate	= {2002/1}
}

@InCollection{	  aitsab96a,
  author	= {O. Aitsab and R. Pyndiah and B. Solaiman},
  title		= {Joint optimization of multi-dimensional {SOFM} codebooks
		  with QAM modulations for vector quantized image
		  transmission},
  booktitle	= {Proceedings IWISPO '96. Third International Workshop on
		  Image and Signal Processing on the Theme of Advances in
		  Computational Intelligence},
  publisher	= {Elsevier},
  year		= {1996},
  editor	= {B. G. Mertzios and P. Liatsis},
  address	= {Amsterdam, Netherlands},
  pages		= {3--6},
  dbinsdate	= {oldtimer}
}

@Article{	  ajjimarangsee89a,
  author	= {P. Ajjimarangsee and T. L. Huntsberger},
  title		= {Neural network model for fusion of visible and infrared
		  sensor outputs},
  journal	= {Proc. SPIE---The International Society for Optical
		  Engineering},
  year		= {1989},
  volume	= {1003},
  pages		= {153--160},
  publisher	= {SPIE},
  address	= {Bellingham, WA},
  note		= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ajjimarangsee89b,
  author	= {P. Ajjimarangsee and T. L. Huntsberger},
  title		= {Unsupervised pattern recognition using parallel
		  \mbox{self-organizing} feature maps},
  booktitle	= {Proc. 4th Conference on Hypercubes, Concurrent Computers
		  and Applications},
  year		= {1989},
  volume	= {II},
  pages		= {1093--1096},
  organization	= {D. O. E. ; US Air Force; NASA},
  publisher	= {Golden Gate Enterprises},
  address	= {Los Altos, CA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  akingbehin90a,
  author	= {K. Akingbehin and K. Khorasani and A. Shaout},
  title		= {Alternative models for neural computing},
  booktitle	= {Proc. 2nd IASTED Int. Symp. Expert Systems and Neural
		  Networks},
  year		= {1990},
  editor	= {M. H. Hamza},
  pages		= {66--69},
  organization	= {IASTED},
  publisher	= {Acta Press},
  address	= {Anaheim, CA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  al-sulaiman93a,
  author	= {M. A. Al-Sulaiman and S. I. Ahson and M. I. Al-Kanhal},
  title		= {Construction of {A}rabic Phoneme Maps using {L}earning
		  {V}ector {Q}uantization},
  booktitle	= {Proc. WCNN'93, World Congress on Neural Networks},
  year		= {1993},
  volume	= {IV},
  pages		= {84--90},
  organization	= {{INNS}},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  alahakoon00a,
  author	= {Alahakoon, Damminda and Halgamuge, Saman K. and
		  Srinivasan, Bala},
  title		= {Dynamic self-organizing maps with controlled growth for
		  knowledge discovery},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {2000},
  volume	= {11},
  number	= {3},
  month		= {May},
  pages		= {601--614},
  organization	= {Monash Univ},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {The growing self-organizing map (GSOM) has been presented
		  as an extended version of the self-organizing map (SOM),
		  which has significant advantages for knowledge discovery
		  applications. In this paper, the GSOM algorithm is
		  presented in detail and the effect of a spread factor,
		  which can be used to measure and control the spread of the
		  GSOM, is investigated. The spread factor is independent of
		  the dimensionality of the data and as such can be used as a
		  controlling measure for generating maps with different
		  dimensionality, which can then be compared and analyzed
		  with better accuracy. The spread factor is also presented
		  as a method of achieving hierarchical clustering of a data
		  set with the GSOM. Such hierarchical clustering allows the
		  data analyst to identify significant and interesting
		  clusters at a higher level of the hierarchy, and as such
		  continue with finer clustering of only the interesting
		  clusters. Therefore, only a small map is created in the
		  beginning with a low spread factor, which can be generated
		  for even a very large data set. Further analysis is
		  conducted on selected sections of the data and as such of
		  smaller volume. Therefore, this method facilitates the
		  analysis of even very large data sets.},
  dbinsdate	= {2002/1}
}

@Article{	  alahakoon98a,
  author	= {Alahakoon, D. and Halgamuge, S. K. and Srinivasan, B.},
  title		= {Self growing cluster development approach to data mining},
  journal	= {IEEE International Conference on Systems, Man, and
		  Cybernetics},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1998},
  number	= {},
  volume	= {3},
  pages		= {2901--2906},
  abstract	= {We describe a data analysis method using a structure
		  adapting neural network with two additional layers. The
		  neural network used is an extended version of a self
		  organizing feature map which can adapt it's structure to
		  better represent the clusters in data. Once the clusters
		  are identified, we use two additional layers on the feature
		  map to analyze the clusters and the representation of
		  attributes in the clusters. Simulations and initial results
		  with two simple benchmark data sets are also described.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alahakoon98b,
  author	= {Alahakoon, D. and Halgamuge, S. K.},
  title		= {Knowledge discovery with supervised and unsupervised self
		  evolving neural networks},
  booktitle	= {Proceedings of the 5th International Conference on Soft
		  Computing and Information/Intelligent Systems.
		  Methodologies for the Conception, Design and Application of
		  Soft Computing},
  publisher	= {World Scientific},
  address	= {Singapore},
  year		= {1998},
  volume	= {2},
  pages		= {907--10},
  abstract	= {We describe two extended algorithms useful in knowledge
		  discovery and data mining. Firstly the self organising
		  feature map (SOFM) extended by acquiring the ability of
		  self growing nodes is described. The second method is a
		  generalised form of radial basis function (RBF) networks
		  with the capability of expanding the hidden layer
		  automatically. The main contribution of this paper is to
		  highlight the fact that the extension of the existing fixed
		  structure neural networks into self evolving neural
		  networks, convert them into useful tools for data mining
		  and knowledge discovery.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alahakoon99a,
  author	= {Alahakoon, L. D. and Halgamuge, S. K. and Srinivasan, B.},
  title		= {A self generating neural architecture for data analysis},
  booktitle	= {IJCNN'99. International Joint Conference on Neural
		  Networks. Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {5},
  pages		= {3548--51},
  abstract	= {Supervised and unsupervised self generating neural network
		  architectures have been used in the recent past. Our
		  previous work (1998) has described an unsupervised self
		  generating feature map, called the growing self organising
		  map (GSOM). In this paper we describe some extensions to
		  the GSOM such that it could be used to map and analyse more
		  realistic data sets.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alahakoon99b,
  author	= {Alahakoon, D. and Halgamuge, S. K. and Srinivasan, B.},
  title		= {Data mining with self generating neuro-fuzzy classifiers},
  booktitle	= {FUZZ-IEEE'99. 1999 IEEE International Fuzzy Systems.
		  Conference Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {2},
  pages		= {1096--101},
  abstract	= {Self generating neural networks have been presented as a
		  better alternative to fixed structure networks in data
		  mining applications. It has also been shown that the
		  nearest prototype classifier is functionally equivalent to
		  an alternative fuzzy classifier model. Several supervised
		  neural networks have been developed to generate nearest
		  prototypes which can be converted to fuzzy rules. We
		  present an extended version of our growing self-organising
		  map (GSOM) model which can also be used to identify nearest
		  prototypes for generating fuzzy rules.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alander90a,
  author	= {Jarmo T. Alander and Antti Autere and Lasse
		  Holmstr{\"{o}}m and Peter Holmstr{\"{o}}m and Ari
		  H{\"{a}}m{\"{a}}l{\"{a}}inen and Juha Tuominen},
  title		= {Surface Type Recognition by a Hair Sensor},
  booktitle	= {Communication Control and Signal Processing},
  editor	= {E. Arikan},
  publisher	= {Elsevier},
  address	= {Amsterdam, Netherlands},
  year		= {1990},
  pages		= {1757--1764},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alander91a,
  author	= {Jarmo T. Alander and Matti Frisk and Lasse Holmstr{\"{o}}m
		  and Ari H{\"{a}}m{\"{a}}l{\"{a}}inen and Juha Tuominen},
  title		= {Process Error Detection using Self-Organizing Feature
		  Maps},
  booktitle	= {Artificial Neural Networks},
  year		= {1991},
  editor	= {T. Kohonen and K. M{\"{a}}kisara and O. Simula and J.
		  Kangas},
  volume	= {II},
  pages		= {1229--1232},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  month		= {},
  dbinsdate	= {oldtimer}
}

@TechReport{	  alander91b,
  author	= {Jarmo T. Alander and Matti Frisk and Lasse Holmstr{\"{o}}m
		  and Ari H{\"{a}}m{\"{a}}l{\"{a}}inen and Juha Tuominen},
  title		= {Process Error Detection Using \mbox{Self-organizing}
		  Feature Maps},
  institution	= {Rolf Nevanlinna Institute},
  address	= {Helsinki, Finland},
  type		= {Res. Reports},
  number	= {A5},
  year		= {1991},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alba01a,
  author	= {Alba, J. L. and Pujol, A. and Villanueva, J. J.},
  title		= {Separating geometry from texture to improve face
		  analysis},
  booktitle	= {IEEE International Conference on Image Processing},
  year		= {2001},
  editor	= {},
  volume	= {2},
  pages		= {673--676},
  organization	= {},
  publisher	= {},
  address	= {},
  abstract	= {This article studies the effect of preprocessing a
		  classical PCA decomposition using a modified Self
		  Organizing Map (SOM) in order to find shape clusters to
		  improve the texture analysis by means of a pool of PCAs. In
		  most successfully view-based recognition systems, shape and
		  texture are jointly used to statistically model a linear or
		  piece-wise linear subspace that optimally explains the face
		  space for a specific database. Our work is aimed to
		  separate the influence that variance in face shape stamps
		  on the set of eigenfaces in the classical PCA
		  decomposition. A set of experiments show the reliability of
		  this new system.},
  dbinsdate	= {2002/1}
}

@Article{	  albeverio97a,
  author	= {S. Albeverio and N. Kruger and B. Tirozzi},
  title		= {An extended {K}ohonen phonetic map},
  journal	= {Mathematical and Computer Modelling},
  year		= {1997},
  volume	= {25},
  number	= {2},
  pages		= {69--73},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  albrecht94a,
  author	= {Albrecht, T. and Matz, G. and Hunte, T. and Hildemann, J.
		  },
  title		= {An intelligent gas sensor system for the identification of
		  hazardous airborne compounds using an array of
		  semiconductor gas sensors and {K}ohonen feature map neural
		  networks},
  booktitle	= {Second Internatinal Conference on 'Intelligent Systems
		  Engineering'},
  year		= {1994},
  pages		= {130--7},
  organization	= {Tech. Univ. Hamburg-Harburg, Germany},
  publisher	= {IEE},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@Article{	  alder90a,
  author	= {Michael Alder and Roberto Togneri and Edmund Lai and
		  Yianni Attikiouzel},
  title		= {{K}ohonen's algorithm for the numerical parametrisation of
		  manifolds},
  journal	= {Pattern Recognition Letters},
  year		= {1990},
  volume	= {11},
  pages		= {313--319},
  dbinsdate	= {oldtimer}
}

@Article{	  alder91a,
  author	= {M. D. Alder and R. Togneri and Y. Attikiouzel},
  title		= {Dimension of the speech space},
  journal	= {IEE Proc. I [Communications, Speech and Vision]},
  year		= {1991},
  volume	= {138},
  number	= {3},
  pages		= {207--214},
  month		= {June},
  dbinsdate	= {oldtimer}
}

@Article{	  alhoniemi00a,
  author	= {Alhoniemi, Esa},
  title		= {Analysis of pulping data using the Self-Organizing Map},
  journal	= {TAPPI Journal},
  year		= {2000},
  volume	= {83},
  number	= {7},
  month		= {Jul},
  pages		= {66},
  organization	= {Helsinki Univ of Technology},
  publisher	= {TAPPI Press},
  address	= {Norcross, GA},
  abstract	= {Analysis of process data makes it possible to obtain
		  useful information from processes or phenomena that are
		  analytically difficult to deal with. To demonstrate the
		  possibilities of this kind of approach, an analysis of the
		  faults that occurred in a continuous digester is presented.
		  The process data are fed to an artificial neural network,
		  the self organizing map (SOM), which is used to form visual
		  presentations of the data. By interpreting these
		  visualizations, the reason for the faults can be
		  determined.},
  dbinsdate	= {2002/1}
}

@InCollection{	  alhoniemi96a,
  author	= {E. Alhoniemi and O. Simula and J. Vesanto},
  title		= {Monitoring and modeling of complex processes using the
		  \mbox{self-organizing} map},
  booktitle	= {Progress in Neural Information Processing. Proceedings of
		  the International Conference on Neural Information
		  Processing},
  publisher	= {Springer-Verlag},
  year		= {1996},
  volume	= {2},
  editor	= {S. -I. Amari and L. Xu and L. -W. Chan and I. King and K.
		  -S. Leung},
  address	= {Singapore},
  pages		= {1169--74},
  dbinsdate	= {oldtimer}
}

@Article{	  alhoniemi99a,
  author	= {Alhoniemi, Esa and Hollm\'en, Jaakko and Simula, Olli and
		  Vesanto, Juha},
  title		= {Process monitoring and modeling using the
		  \mbox{self-organizing} map},
  journal	= {Integrated Computer-Aided Engineering},
  year		= {1999},
  number	= {1},
  volume	= {6},
  pages		= {3--14},
  abstract	= {The Self-Organizing Map (SOM) is a powerful neural network
		  method for analysis and visualization of high-dimensional
		  data. It maps nonlinear statistical dependencies between
		  high-dimensional measurement data into simple geometric
		  relationships on a usually two-dimensional grid. The
		  mapping roughly preserves the most important topological
		  and metric relationships of the original data elements and,
		  thus, inherently clusters the data. The need for
		  visualization and clustering occurs, for instance, in the
		  analysis of various engineering problems. In this paper,
		  the {SOM} has been applied in monitoring and modeling of
		  complex industrial processes. Case studies, including pulp
		  process, steel production, and paper industry axe
		  described.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alhoniemi99b,
  author	= {Alhoniemi, E. and Himberg, J. and Vesanto, J.},
  title		= {Probabilistic measures for responses of self-organizing
		  map units},
  booktitle	= {Proceedings of the International ICSC Congress on
		  Computational Intelligence Methods and Applications. ICSC
		  Academic Press, Zurich, Switzerland},
  year		= {1999},
  volume	= {},
  pages		= {},
  abstract	= {The self-organizing map (SOM) is a widely used data
		  visualization tool in engineering applications. The
		  algorithm performs a nonlinear mapping from a
		  high-dimensional data space to a low-dimensional space,
		  which is typically a two-dimensional, rectangular grid.
		  This makes it possible to present multidimensional data in
		  two dimensions. Often the model vectors of the SOM and a
		  new data sample need to be compared. The SOM, however,
		  gives no probability measures to determine if the sample
		  belongs to data sets determined by map units. For this
		  purpose a modified batch version of reduced kernel density
		  estimator (RXDE) was tested. The results were compared with
		  Gaussian mixture model (GMM) and S-Map.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  alhumaidi96a,
  author	= {S. M. Alhumaidi and W. L. Jones and Jun-Dong Park and S.
		  Ferguson and M. H. Thursby and S. H. Yueh},
  title		= {A neural network sea ice edge classifier for the {NASA}
		  Scatterometer},
  booktitle	= {IGARSS '96. 1996 International Geoscience and Remote
		  Sensing Symposium. Remote Sensing for a Sustainable
		  Future},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {3},
  editor	= {T. I. Stein},
  address	= {New York, NY, USA},
  pages		= {1526--8},
  dbinsdate	= {oldtimer}
}

@Article{	  ali93a,
  author	= {Ali, K. S. },
  title		= {Self learning for autonomous systems},
  journal	= {Computers \& Industrial Engineering},
  year		= {1993},
  volume	= {25},
  pages		= {401--4},
  month		= {Sept},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InCollection{	  alici96a,
  author	= {Y. Alici},
  title		= {Neural networks in corporate failure prediction: the UK
		  experience},
  booktitle	= {Neural Networks in Financial Engineering. Proceedings of
		  the Third International Conference on Neural Networks in
		  the Capital Markets},
  publisher	= {World Scientific},
  year		= {1996},
  editor	= {A. -P. N. Refenes and Y. Abu-Mostafa and J. Moody and A.
		  Weigend},
  address	= {Singapore},
  pages		= {393--406},
  dbinsdate	= {oldtimer}
}

@Article{	  alici96b,
  author	= {Alici, Yurt},
  title		= {Corporate solvency map through \mbox{self-organising}
		  neural networks},
  journal	= {IEEE IAFE Conference on Computational Intelligence for
		  Financial Engineering (CIFEr)},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1996},
  number	= {},
  volume	= {},
  pages		= {286--292},
  abstract	= {Corporate failure prediction has been used in the
		  application of both parametric classical classification and
		  non-parametric artificial neural network techniques.
		  Although discriminant and logistic regression analysis have
		  been accepted as standard pattern recognition devices,
		  different kinds of neural network technology have recently
		  demonstrated promising outcomes, in terms of accuracy, when
		  compared with results from classical pattern recognition
		  techniques. Most of the neural net studies in corporate
		  failure prediction have centred on implementing a large
		  variety of supervised learning algorithms. Considering
		  stochastic properties of financial ratios due to creative
		  accounting practices, different accounting policies and
		  deviant patterns of so-called healthy companies, little
		  work has been conducted in identifying different patterns
		  of both failed and solvent firms. Therefore, the purpose of
		  this study is to extract solvency maps of UK listed
		  manufacturing firms, by employing self-organising maps. The
		  results obtained from this research indicate that there is
		  marked difference between failed and non-failed firms in
		  terms of financial characteristics although different
		  financial structures exist amongst both bankrupt and
		  solvent companies.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  alirezaie95a,
  author	= {J. Alirezaie and M. E. Jernigan and C. Nahmias},
  title		= {Neural network based segmentation of magnetic resonance
		  images of the brain},
  booktitle	= {1995 IEEE Nuclear Science Symposium and Medical Imaging
		  Conference Record},
  publisher	= {IEEE},
  year		= {1995},
  volume	= {3},
  editor	= {P. A. Moonier},
  address	= {New York, NY, USA},
  pages		= {1397--401},
  dbinsdate	= {oldtimer}
}

@Article{	  alirezaie97a,
  author	= {J. Alirezaie and M. E. Jernigan and C. Nahmias},
  title		= {Neural network-based segmentation of magnetic resonance
		  images of the brain},
  journal	= {IEEE Transactions on Nuclear Science},
  year		= {1997},
  volume	= {44},
  number	= {2},
  pages		= {194--8},
  note		= {(1995 Nuclear Science Symposium and Medical Imaging
		  (NSS/MIC) Conf. Date: 21--28 Oct. 1995 Conf. Loc: San
		  Francisco, CA, USA)},
  dbinsdate	= {oldtimer}
}

@Article{	  alirezaie97b,
  author	= {J. Alirezaie and M. E. Jernigan and C. Nahmias},
  title		= {Automatic segmentation of {MR} images using
		  \mbox{self-organizing} feature mapping and neural networks},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1997},
  volume	= {3034},
  number	= {pt. 1--2},
  pages		= {138--49},
  note		= {(Medical Imaging 1997: Image Processing Conf. Date: 25--28
		  Feb. 1997 Conf. Loc: Newport Beach, CA, USA Conf. Sponsor:
		  SPIE)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  allinson89a,
  author	= {Nigel M. Allinson and Martin J. Johnson and Kevin J.
		  Moon},
  title		= {Digital realisation of {S}elf-{O}rganising {M}aps},
  booktitle	= {Advances in Neural Information Processing Systems I},
  year		= {1989},
  editor	= {David S. Touretzky},
  pages		= {728--738},
  publisher	= {Morgan Kaufmann},
  address	= {San Mateo, CA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  allinson89b,
  author	= {N. M. Allinson and M. T. Brown and M. J. Johnson},
  title		= {{{0,1}N} space \mbox{self-organising} feature
		  maps---extensions and hardware},
  booktitle	= {IEE International Conference on Artificial Neural
		  Networks, Publication 313},
  year		= {1989},
  pages		= {261--264},
  publisher	= {IEE},
  address	= {London, UK},
  annote	= {},
  dbinsdate	= {oldtimer}
}

@InCollection{	  allinson89c,
  author	= {N. M. Allinson and M. J. Johnson},
  title		= {Realisation of self organising neural maps in {{0,1}N}
		  space},
  booktitle	= {{New Developments in Neural Computing}},
  publisher	= {Adam-Hilger},
  year		= {1989},
  editor	= {J. G. Taylor and C. L T. Mannion},
  chapter	= {},
  pages		= {79--86},
  address	= {Bristol, UK},
  dbinsdate	= {oldtimer}
}

@Article{	  allinson92a,
  author	= {N. M. Allinson and A. W. Ellis},
  title		= {Face recognition: combining cognitive psychology and image
		  engineering},
  journal	= {IEE Electronics and Communication J. },
  year		= {1992},
  volume	= {4},
  number	= {},
  pages		= {291--300},
  annotate	= {},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  allinson92b,
  author	= {N. M. Allinson and M. J. Johnson},
  title		= {Application of \mbox{self-organising} digital neural
		  networks in attentive vision systems},
  booktitle	= {Proc. Fourth Int. IEE Conference on Image Processing and
		  its Applications, Maastricht, Netherlands},
  year		= {1992},
  pages		= {193--196},
  annote	= {},
  dbinsdate	= {oldtimer}
}

@InCollection{	  allinson92c,
  author	= {N. M. Allinson},
  title		= {Self-organising neural maps and their applications},
  booktitle	= {Theory and Applications of Neural Networks},
  publisher	= {Springer},
  year		= {1992},
  editor	= {J. G. Taylor and C. L. T. Mannion},
  chapter	= {},
  pages		= {101--120},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  allinson93a,
  author	= {N. M. Allinson and H. Yin},
  title		= {Comparison of {K}ohonen \mbox{self-organising} maps and
		  {K}alman filtering},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {},
  dbinsdate	= {oldtimer}
}

@InCollection{	  allinson98a,
  author	= {N. M. Allinson and Hujun Yin},
  title		= {Interactive and semantic data visualisation using
		  \mbox{self-organising} maps},
  booktitle	= {IEE Colloquium Neural Networks in Interactive Multimedia
		  Systems},
  publisher	= {IEE},
  year		= {1998},
  address	= {London, UK},
  pages		= {1/1--10},
  dbinsdate	= {oldtimer}
}

@InCollection{	  allinson99a,
  author	= {N. M. Allinson and H. Yin},
  title		= {Self-Organising Maps for pattern recognition},
  booktitle	= {Kohonen Maps},
  pages		= {111--20},
  publisher	= {Elsevier},
  year		= {1999},
  editor	= {Oja, E. and Kaski, S.},
  address	= {Amsterdam},
  annote	= {},
  dbinsdate	= {2002/1}
}

@Proceedings{	  allison01a,
  title		= {Advances in Self-Organising Maps},
  year		= {2001},
  editor	= {Nigel Allinson and Hujun Yin and Lesley Allinson and Jon
		  Slack},
  publisher	= {Springer},
  dbinsdate	= {2002/1}
}

@Article{	  almajali01a,
  author	= {Almajali, S. and Sharieh, A. and Qutiashat, M.},
  title		= {Arabic speech recognition using {SOM}-{LVQ} neural
		  networks},
  journal	= {Advances-in-Modelling-\&-Analysis-B:-Signals,-Information,-Patterns,-Data-Acquisition,-Transmission,-Processing,-Classification}
		  ,
  year		= {2001},
  volume	= {44},
  pages		= {1--15},
  abstract	= {In this research, an Arabic speech recognition system
		  (ASRS) has been developed using self-organized map (SOM)
		  neural networks, which adopts learning vector quantization
		  (LVQ). In this system, Arabic speech recognition was
		  successfully developed and tested for both phonemes and a
		  set of words. To increase recognition accuracy, the 38
		  Arabic phonemes were arranged into a set of classes based
		  on linguistics and acoustics principles. For each class of
		  phonemes, a neural network that performs the recognition
		  task for that class was built. For vowel-semivowel,
		  nasal-glide, and subset of unvoiced phoneme classes, the
		  recognition generalization accuracy (GA) of tests reached
		  around 92%, 96%, 98%, respectively. The ASRS produced GA
		  equals to 80.49% for all Arabic phonemes, 86.9% for 33
		  phonemes. 89.5% for selected 25 phonemes, and 96% for
		  selected 20 phonemes. A set of words that found in the
		  Arabic menus of Microsoft Word package was selected to be
		  tested by the ASRS. For each menu, a separate neural
		  network/map has been created. The GA using SOM for the nine
		  menus was around 88%. The GA after 1500 LVQ steps was
		  around 91% and after 3000 LVQ steps was around 92%.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  alpaydin93a,
  author	= {R. Alpaydin and U. {\"{U}}nl{\"{u}}akin and F.
		  G{\"{u}}rgen and E. Alpaydin},
  title		= {Comparing Distributed and Local Neural Classifiers for the
		  Recognition of Japanese Phonemes},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {I},
  pages		= {239--242},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  alrashdan00a,
  author	= {Alrashdan, A. and Motavalli, S. and Fallahi, B.},
  title		= {Automatic segmentation of digitized data for reverse
		  engineering applications},
  journal	= {IIE Transactions},
  year		= {2000},
  volume	= {32},
  pages		= {59--69},
  abstract	= {Reverse engineering is the process of developing a CAD
		  model and a manufacturing database for an existing part.
		  The paper is about the automatic segmentation of 3D
		  digitized data for such applications. This is achieved
		  using a combination of region and edge based approaches. It
		  is assumed that the part surface contains planar as well as
		  curved surfaces that are embedded in a base surface. The
		  part surface should be visible to a single scanning probe
		  (2 1/2 D object). Neural net algorithms are developed. A
		  backpropagation net is used to segment part surfaces into
		  surface primitives homogenous in their intrinsic
		  differential geometric properties. The method is based on
		  the computation of Gaussian and mean curvatures, obtained
		  by locally approximating the object surface using quadratic
		  polynomials. They are used as input to the neural net which
		  outputs an initial region-based segmentation in the form of
		  a curvature sign map. An edge based segmentation is also
		  performed using the partial derivatives of depth values.
		  Here, the output of the Laplacian operator and the unit
		  surface normal are computed and used as input to a
		  self-organized mapping (SOM) net which is used to find the
		  edge points. The combination of approaches segments the
		  data into primitive surface regions. The uniqueness of our
		  approach is in automatic calculation of the threshold level
		  for segmentation, and on the adaptability of the method to
		  various noise levels in the digitized data. The developed
		  algorithms and sample results are described in the paper.},
  dbinsdate	= {oldtimer}
}

@Article{	  altinel00a,
  author	= {Altinel, I. K. and Aras, N. and Oommen, B. J.},
  title		= {Fast, efficient and accurate solutions to the Hamiltonian
		  path problem using neural approaches},
  journal	= {COMPUTERS \& OPERATIONS RESEARCH},
  year		= {2000},
  volume	= {27},
  number	= {5},
  month		= {APR},
  pages		= {461--494},
  abstract	= {Unlike its cousin, the Euclidean Traveling Salesman
		  Problem (TSP), to the best of our knowledge, there has been
		  no documented all-neural solution to the Euclidean
		  Hamiltonian Path Problem (HPP). The reason for this is the
		  fact that the heuristics which map the cities onto the
		  neurons "lose their credibility" because the underlying
		  cyclic property of the order of the neurons used in the TSP
		  is lost in the HPP. In this paper we present three neural
		  solutions to the HPP. The first of these, GSOM_HPP, is a
		  generalization of Kohonen's self-organizing map (SOM) as
		  modified by Angeniol et al. (Neural Networks
		  1988;1:289--93). The second and third methods use the
		  recently-introduced self-organizing neural network, the
		  Kohonen Network Incorporating Explicit Statistics (KNIES)
		  (Oommen et al., Proceedings of WIRN/VIETRI-98, the Tenth
		  Italian Workshop on Neural Nets, Vietri Sul Mare, Italy,
		  May 1998. p. 273--282). The primary difference between
		  KNIES and Kohonen's SOM is the fact that unlike SOM, every
		  iteration in the training phase includes two distinct
		  modules---the attracting module and the dispersing module.
		  As a result of SOM and the dispersing module introduced in
		  KNIES the neurons individually find their places both
		  statistically and topologically, and also collectively
		  maintain their mean to be the mean of the data points which
		  they represent. The new philosophy, which has previously
		  (Oommen et al. Proceedings of WIRN/VIETRI-98, the Tenth
		  Italian Workshop on Neural Nets, Vietri Sul Mare, Italy,
		  May 1998. p. 273--282) been used to effectively solve the
		  Euclidean Traveling Salesman Problem (TSP), is now extended
		  to solve the Euclidean Hamiltonian Path (HPP). These
		  algorithms which are the first all-neural solutions to the
		  HPP, have also been rigorously tested. Experimental results
		  for problems obtained by modifying selected instances from
		  the traveling salesman problem library (TSPLIB) (Reinett.
		  ORSA Journal on Computing 1991;3:376--84) for the HPP
		  indicate that they are both accurate and efficient. Apart
		  from the computational results presented, the paper also
		  contains a systematic strategy by which the quality of any
		  HPP algorithm can be quantified. Scope and purpose Over the
		  past two decades an enormous amount of work has been done
		  in designing neural networks (NNs) which utilize a variety
		  of learning principles. There are many works in the
		  literature that are noteworthy in the context, we list only
		  a few like CS,621, which describe the various families of
		  NNs, and how their learning compares to underlying
		  biological learning models. In this paper we concentrate
		  our attention on Kohonen's Self-Organizing Map (SOM) [21].
		  The SOM has een used in solving certain optimization
		  problems such as the Euclidean Traveling Salesman Problem
		  [3,4] which has been one of the oldest "hard nuts" of
		  Operations Research and Mathematical Programming. Any
		  algorithm devised to solve the TSP tries to answer the
		  following question: Given a set of N cities and distances
		  for each pair of cities what is the shortest tour that
		  visits each city exactly once? The beauty of the SOM is the
		  fact that the individual neurons adaptively tend to learn
		  the properties of the underlying distribution of the space
		  in which they operate. Additionally, they also tend to
		  learn their places topologically. This feature is
		  particularly important for problems which involve two and
		  three-dimensional physical spaces, and is indeed, the
		  principal motivation for the SOM being used in solving the
		  TSP [7,8]. More recently, we have added to this collection
		  of methods a scheme which uses the recently introduced
		  self-organizing neural network, the Kohonen Network
		  Incorporating Explicit Statistics (KNIES) [9]. The primary
		  difference between KNIES and the SOM is that, in KNIES, the
		  neurons not only individually find their places
		  statistically and topologically, but also collectively
		  maintain their mean to be the mean of the data points which
		  they represent. The Euclidean Hamiltonian Path Problem
		  (HPP) is closely related to the TSP: Given a set {X-i:1
		  less than or equal to i less than or equal to N} of cities
		  with starting and terminal cities X-s and X-t,
		  respectively, and distances for each pair of cities, what
		  is the shortest path that starts at X-s, terminates at X-t,
		  respectively, and visits each city exactly once? There are
		  only a few independent solutions to the HPP; indeed most
		  solutions utilize the underlying solution to the TSP. In
		  this paper we try to solve the HPP without resorting to an
		  underlying TSP solution method, and adapt our new NN
		  methods, KNIES. Our new neural algorithms are the first
		  all-neural solutions to the HPP to our knowledge, and
		  experimental results indicate that they are accurate and
		  efficient. },
  dbinsdate	= {2002/1}
}

@InProceedings{	  alvarez94a,
  author	= {M. Alvarez and A. Varfis},
  title		= {Decoding Functions for {K}ohonen Maps},
  booktitle	= {Proc. ESANN'94, European Symp. on Artificial Neural
		  Networks},
  year		= {1994},
  publisher	= {D facto conference services},
  address	= {Brussels, Belgium},
  editor	= {M. Verleysen},
  pages		= {245--250},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alvarez95a,
  author	= {M. Alvarez and J. -M. Auger and A. Varfis},
  title		= {On Self-Organised Regression Curves},
  booktitle	= {Proc. ICANN'95, International Conference on Artificial
		  Neural Networks},
  year		= {1995},
  editor	= {F. Fogelman-Souli{\'{e}} and P. Gallinari},
  volume	= {II},
  pages		= {21--26},
  publisher	= {EC2},
  address	= {Nanterre, France},
  dbinsdate	= {oldtimer}
}

@InCollection{	  alvarez96a,
  author	= {G. Alvarez and S. Guevara and C. Piedrahita and O.
		  Tapias},
  title		= {The application of fuzzy {K}ohonen neural networks to
		  trace editing in seismic data processing},
  booktitle	= {Fourth European Congress on Intelligent Techniques and
		  Soft Computing Proceedings, EUFIT '96},
  publisher	= {Verlag Mainz},
  year		= {1996},
  volume	= {2},
  address	= {Aachen, Germany},
  pages		= {1146--50},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  alves_de_barros93a,
  author	= {Marcelo {Alves de Barros} and Mohamed Akil and Ren{\'{e}}
		  Natowicz},
  title		= {A Reconfigurable Architecture for Real Time Segmentation
		  of Image Sequences using Self-Organizing Feature Maps},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {I},
  pages		= {197--202},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  amarasingham98a,
  author	= {A. Amarasingham and W. B. Levy},
  title		= {Predicting the Distribution of Synaptic Strengths and Cell
		  Firing Correlations in a Self Organizing, Sequence
		  Prediction Model},
  journal	= {Neural Computation},
  volume	= {10},
  pages		= {25--57},
  year		= {1998},
  dbinsdate	= {oldtimer}
}

@InCollection{	  amari89a,
  author	= {{S. -i. } Amari},
  title		= {Dynamical Study of the Formation of Cortical Maps},
  booktitle	= {Dynamic Interactions in Neural Networks: Models and Data},
  publisher	= {Springer},
  address	= {Berlin, Heidelberg},
  year		= 1989,
  editor	= {M. A. Arbib and S. -i. Amari},
  pages		= {15--34},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ambroise95a,
  author	= {Cristophe Ambroise and G{\'{e}}rard Govaert},
  title		= {Self-Organization for {G}aussian Parsimonious Clustering},
  booktitle	= {Proc. ICANN'95, International Conference on Artificial
		  Neural Networks},
  year		= {1995},
  editor	= {F. Fogelman-Souli{\'{e}} and P. Gallinari},
  volume	= {I},
  pages		= {425--430},
  publisher	= {EC2},
  address	= {Nanterre, France},
  dbinsdate	= {oldtimer}
}

@InCollection{	  ambuhl97a,
  author	= {J. Amb{\"u}hl and D. Cattani and P. Eckert},
  title		= {Classification of Meteorological Patterns},
  booktitle	= {Proc. ICANN'97, 7th International Conference on Artificial
		  Neural Networks},
  publisher	= {Springer},
  year		= 1997,
  volume	= 1327,
  series	= {Lecture Notes in Computer Science},
  address	= {Berlin},
  pages		= {1119--1124},
  dbinsdate	= {oldtimer}
}

@Article{	  amerijckx98a,
  author	= {Amerijckx, Christophe and Verleysen, Michel and Thissen,
		  Philippe and Legat, Jean Didier},
  title		= {Image compression by self-organized {K}ohonen map},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1998},
  number	= {3},
  volume	= {9},
  pages		= {503--507},
  abstract	= {This paper presents a compression scheme for digital still
		  images, by using the Kohonen's neural network algorithm,
		  not only for its vector quantization feature, but also for
		  its topological property. This property allows an increase
		  of about 80% for the compression rate. Compared to the JPEG
		  standard, this compression scheme shows better performances
		  (in terms of PSNR) for compression rates higher than 30.},
  dbinsdate	= {oldtimer}
}

@Article{	  amin94a,
  author	= {Shara Amin},
  title		= {A Self-Organized Travelling Salesman},
  journal	= {Neural Computing {\&} Applications},
  year		= {1994},
  volume	= {2},
  number	= {3},
  pages		= {129--133},
  dbinsdate	= {oldtimer}
}

@Article{	  aminian93a,
  author	= {Aminian, K. and Robert, P. and Jequier, E. and Schutz, Y.
		  },
  title		= {Level, downhill and uphill walking identification using
		  neural networks},
  journal	= {Electronics Letters},
  year		= {1993},
  volume	= {29},
  number	= {17},
  pages		= {1563--5},
  month		= {Aug},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anagnostopoulos01a,
  author	= {Anagnostopoulos, C. and Anagnostopoulos, J. and Vergados,
		  D.D. and Kayafas, E. and Loumos, V. and Theodoropoulos,
		  G.},
  title		= {Training a learning vector quantization network for
		  biomedical classification},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2001},
  editor	= {},
  volume	= {4},
  pages		= {2506--2511},
  organization	= {Natl. Tech. Univ. of Athens (NTUA), Electrical and
		  Computer Eng. Dept.},
  publisher	= {},
  address	= {},
  abstract	= {A competitive Learning Vector Quantization (LVQ)
		  Artificial Neural Network (ANN) was trained to identify
		  third stage parasitic strongyle larvae from domestic
		  animals on the basis of quantitative data obtained from
		  processed digital images of larvae. For this reason various
		  novel quantitative features obtained from processed digital
		  images of larvae were tested whether they are variant or
		  invariant to the shape taken by the motile larvae during
		  image recording. A total of 255 images of 57 individual
		  larvae in various shapes belonging to 5 genera were
		  recorded. Following image processing 16 novel features were
		  measured of which 7 were selected as invariant to larva
		  shape. By trial and error two of those novel features
		  "area" and "perimeter" along with the quantitative features
		  used in conventional identification, "overall body length",
		  "width" and "tail of sheath" were used as an effective
		  training data set for the ANN. This ANN coupled with an
		  image analysis facility and a knowledge relational database
		  became the basis for developing a computer-based larva
		  identification system whose overall identification
		  performance was 91.9%. The advantages of this system are
		  its speed and objectivity. The objectivity of the system is
		  based on the fact that it is not subject to inter- and
		  intra-observer variability arising from the user's profile
		  of competency in interpreting subjective and
		  non-quantifiable descriptions. The limitations of the
		  system are that it can not handle raw images but only data
		  extracted from images, its performance depends on the
		  reliability of the input vectors used as training data for
		  the ANN and its use is restricted only in well equipped
		  laboratories due to its requirement for expensive
		  instrumentation.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  anand91a,
  author	= {R. Anand and K. Mehrotra and C. K. Mohan and S. Ranka},
  title		= {Analyzing Images Containing Multiple Sparse Patterns with
		  Neural Networks},
  booktitle	= {Proc. International Joint Conference on Artificial
		  Intelligence (IJCAI)},
  publisher	= {University of Sydney},
  year		= {1991},
  address	= {Sydney, Australia},
  dbinsdate	= {oldtimer}
}

@Article{	  anand93a,
  author	= {R. Anand and K. Mehrotra and C. K. Mohan and S. Ranka},
  title		= {Analyzing Images Containing Multiple Sparse Patterns with
		  Neural Networks},
  journal	= {Pattern Recognition},
  year		= 1993,
  volume	= 26,
  pages		= {1717--1724},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anderson90a,
  author	= {Ove Anderson and Piero Cosi and Paul Dalsgaard},
  title		= {A {SONN}-based Architecture for Automatic Speech
		  Segmentation and Alignment},
  editor	= {Andrea Paoloni},
  pages		= {18--29},
  booktitle	= {Proc. 1st Workshop on Neural Networks and Speech
		  Processing, November 89, Roma},
  address	= {Roma, Italy},
  year		= {1990},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anderson91a,
  author	= {T. R. Anderson. },
  title		= {Speaker independent phoneme recognition with an auditory
		  model and a neural network: a comparison with traditional
		  techniques},
  booktitle	= {Proc. ICASSP-91, International Conference on Acoustics,
		  Speech and Signal Processing},
  year		= {1991},
  volume	= {I},
  pages		= {149--152},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anderson92a,
  author	= {Anderson, T. R. },
  title		= {Phoneme recognition using an auditory model and a
		  recurrent \mbox{self-organizing} neural network},
  booktitle	= {ICASSP-92: 1992 IEEE International Conference on
		  Acoustics, Speech and Signal Processing},
  year		= {1992},
  volume	= {2},
  pages		= {337--40},
  organization	= {Armstrong Lab. , Wright-Patterson AFB, OH, USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anderson94a,
  author	= {Timothy Anderson},
  title		= {Auditory Models with {K}ohonen {SOFM} and {LVQ} for
		  Speaker Independent Phoneme Recognition},
  pages		= {4466--4469},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, speech recognition},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anderson94b,
  author	= {Anderson, T. R. and Patterson, R. D. },
  title		= {Speaker recognition with the auditory image model and
		  \mbox{self-organizing} feature maps: A comparison with
		  traditional techniques},
  booktitle	= {ESCA Workshop on Automatic Speaker Recognition
		  Identification and Verification},
  year		= {1994},
  pages		= {153--6},
  organization	= {Armstrong Lab. , Wright Res. \& Dev. Center,
		  Wright-Patterson AFB, OH, USA},
  publisher	= {IDIAP},
  address	= {Martingny, Switzerland},
  dbinsdate	= {oldtimer}
}

@Article{	  anderson99a,
  author	= {Anderson, B.},
  title		= {{K}ohonen Neural Networks and Language},
  journal	= {Brain and Language},
  year		= {1999},
  volume	= {70},
  number	= {1},
  pages		= {86--94},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ando91a,
  author	= {Akio Ando and Kazuhiko Ozeki},
  title		= {A Multi-Template Learning Algorithm Based on Minimization
		  of Recognition Error Function},
  booktitle	= {Artificial Neural Networks},
  year		= {1991},
  editor	= {Kohonen, Teuvo and M{\"{a}}kisara, Kai and Simula, Olli
		  and Kangas, Jari},
  pages		= {421--426},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@Article{	  andrade97a,
  author	= {Andrade,M. A. and Casari,G. and Sander,C. and Valencia,A.
		  },
  title		= {Classification of Protein Families and Detection of the
		  Determinant Residues with an Improved Self Organizing Map},
  journal	= {Biol. Cyb. },
  year		= {1997},
  pages		= {441--50},
  volume	= {76},
  dbinsdate	= {oldtimer}
}

@Article{	  andrare93a,
  author	= {M. A. Andrare and P. Chac{\'{o}}n and J. J. Merelo and F.
		  Mor{\'{a}}n},
  title		= {Evaluation of Secondary Structure of Proteins from {UV}
		  Circular Dichroism Spectra using an Unsupervised Learning
		  Neural Network},
  journal	= {Protein Engineering},
  year		= {1993},
  volume	= {6},
  number	= {4},
  pages		= {383--390},
  dbinsdate	= {oldtimer}
}

@Article{	  andreou89a,
  author	= {A. G. Andreou and K. A. Boahen},
  title		= {Synthetic neural circuits using current-domain signal
		  representations},
  journal	= {Neural Computation},
  year		= {1989},
  volume	= {1},
  number	= {4},
  pages		= {489--501},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andres92a,
  author	= {Andres, M. and Mallot, H. and Giefing, G. -J. },
  title		= {Selforganization of binocular receptive fields},
  booktitle	= {Artificial Neural Networks, 2. Proceedings of the 1992
		  International Conference (ICANN-92)},
  year		= {1992},
  editor	= {Aleksander, I. },
  volume	= {1},
  pages		= {553--6},
  organization	= {Inst. fur Neuroinf. , Ruhr-Univ. Bochum, Germany},
  publisher	= {Elsevier},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andres94a,
  author	= {Marianne Andres and Oliver Schl{\"{u}}ter and Friederike
		  Spengler and Hubert R. Dinse},
  title		= {A Model of Fast and Reversible Representation Plasticity
		  using {K}ohonen Mapping},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  pages		= {306--309},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, physiological modeling},
  dbinsdate	= {oldtimer}
}

@InCollection{	  andres96a,
  author	= {M. Andres and O. Schluter and F. Spengler and H. R.
		  Dinse},
  title		= {Modification of {K}ohonen's {SOFM} to simulate cortical
		  plasticity induced by coactivation input patterns},
  booktitle	= {Artificial Neural Networks---ICANN 96. 1996 International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1996},
  editor	= {C. {von der Malsburg} and W. {von Seelen} and J. C.
		  Vorbruggen and B. Sendhoff},
  address	= {Berlin, Germany},
  pages		= {421--6},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andreu96a,
  author	= {Andreu, G. and Crespo, A.},
  title		= {Estimating the organization degree for a toroidal
		  \mbox{self-organizing} feature map ({TSOFM})},
  booktitle	= {Solving Engineering Problems with Neural Networks.
		  Proceedings of the International Conference on Engineering
		  Applications of Neural Networks (EANN'96)},
  publisher	= {Systems Engineering Association, Turku, Finland},
  year		= {1996},
  volume	= {2},
  pages		= {649--52},
  abstract	= {Feature maps (SOFM) are an important tool to visualize
		  high-dimensional data as a two-dimensional image. One of
		  the possible application of this network is to image
		  recognition. However, this architecture presents some
		  problems mainly due to the border effects. In this paper a
		  new organization of the feature maps named toroidal
		  self-organizing feature maps ({TSOFM}) is presented. Its
		  main advantage consist on the elimination of the the border
		  effects and, consequently, the increasing of the
		  recognition rate. Another important aspect presented in
		  this paper is the measurement of how well organized is the
		  networks during the training phase. This proposal have been
		  experimented in an industrial application demonstrating the
		  better quality results.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  andreu97a,
  author	= {G. Andreu and A. Crespo and J. M. Valiente},
  title		= {Selecting the Toroidal Self-Organizing Feature Maps
		  ({TSOFM}) Best Organized to Object Recognition},
  booktitle	= {Proceedings of ICNN'97, International Conference on Neural
		  Networks},
  publisher	= {IEEE Service Center},
  year		= 1997,
  address	= {Piscataway, NJ},
  volume	= {II},
  pages		= {1341--1346},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andrew94a,
  author	= {Lachlan L. H. Andrew and M. Palaniswami},
  title		= {A Study on the Effect of Neighbourhood Functions for Noise
		  Robust Vector Quantisers},
  pages		= {4159--4162},
  booktitle	= {Proc. ICNN'94 IEEE International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, vector quantization, noise tolerance},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andrew94b,
  author	= {Lachlan L. H. Andrew},
  title		= {Neuron Splitting for Efficient Feature Map Formation},
  booktitle	= {Proc. ANZIIS'94, Aust. New Zealand Intell. Info. Systems
		  Conference },
  pages		= {10--13},
  year		= 1994,
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {modification},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andrew95a,
  author	= {Lachlan L. H. Andrew and M. Palaniswami},
  title		= {A New Adaptive Image Sequence Coding Scheme Using
		  {K}ohonen's {SOFM}},
  volume	= {IV},
  pages		= {2071--2076},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, vector quantization, image sequence coding},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andrew95b,
  author	= {Colin Andrew and Miroslaw Kubat and Gert Pfurtscheller},
  title		= {Trimming the Inputs of {RBF} Networks},
  booktitle	= {Proc. ESANN'95, European Symp. on Artificial Neural
		  Networks},
  year		= {1995},
  publisher	= {D facto conference services},
  address	= {Brussels, Belgium},
  editor	= {M. Verleysen},
  pages		= {291--296},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andrew96a,
  author	= {Lachlan L. H. Andrew},
  title		= {Neural Networks for Adaptive Image Sequence Vector
		  Quantization},
  pages		= {569--573},
  booktitle	= {Proc. IPCS'6 Int. Picture Coding Symposium},
  year		= {1996},
  annote	= {application, vector quantization, image sequence coding},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  andrianasy95a,
  author	= {Fidimahery Andianasy and Maurice Milgram},
  title		= {A Learning Scheme for On-Line Handwritten Recognition
		  Using Elastic Matching},
  booktitle	= {Proc. EANN'95, Engineering Applications of Artificial
		  Neural Networks},
  year		= {1995},
  pages		= {61--65},
  organization	= {Finnish Artificial Intelligence Society},
  dbinsdate	= {oldtimer}
}

@Article{	  angelakis01a,
  author	= {Angelakis, C. and Loukis, E. N. and Pouliezos, A. D. and
		  Stavrakakis, G.S.},
  title		= {A neural network-based method for gas turbine blading
		  fault diagnosis},
  journal	= {International Journal of Modelling and Simulation},
  year		= {2001},
  volume	= {21},
  number	= {1},
  month		= {},
  pages		= {51--60},
  organization	= {Technical University of Crete, Department of Electronic
		  Engineering},
  publisher	= {},
  address	= {},
  abstract	= {In this paper artificial neural networks are used with
		  promising results in a critical, and at the same time, very
		  difficult problem concerning the diagnosis of gas turbine
		  blading faults. Neural network-based fault diagnosis is
		  treated as a pattern recognition problem, based on
		  measurements and feature selection. Emphasis is given to
		  the design of the appropriate neural network architecture
		  and the selection of the appropriate measuring instruments,
		  which are of critical importance for achieving good
		  performance (high success rates and generalization
		  capabilities). Initially the performance of the classical
		  neural network architectures, namely MultiLayer Perceptron
		  (MLP), Learning Vector Quantization (LVQ), Modular
		  MultiLayer Perceptron and Radial Basis Function (RBF), are
		  investigated for this problem. The implemented neural
		  network structures are trained to classify faulty and
		  healthy patterns coming from twelve different measuring
		  instruments. The performance of the above neural network
		  structures is investigated, and the diagnostic capabilities
		  of the measuring instruments are examined. Next, in order
		  to improve the generalization capabilities, which are
		  critical for the specific diagnostic problem, a new
		  multinet architecture is developed, based on the idea of
		  'majority rule' decision. Compared with the classical
		  architectures, this new multinet architecture is
		  characterized by higher generalization capabilities and
		  robustness. A first approach to the design of the
		  appropriate multinet architecture and the selection of the
		  appropriate measuring instruments, in order to provide the
		  basis of a high-performance automated diagnostic system, is
		  proposed. The conclusions derived are of general interest
		  and applicability.},
  dbinsdate	= {2002/1}
}

@Article{	  angeniol88a,
  author	= {B. Ang{\`{e}}niol and G. D. L. C. Vaubois and J. Y. L.
		  Texier},
  title		= {Self-organizing feature maps and the {T}ravelling
		  {S}alesman {P}roblem},
  journal	= {Neural Networks},
  year		= {1988},
  volume	= {1},
  number	= {4},
  pages		= {289--293},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  anguita95a,
  author	= {Davide Anguita and Filippo Passaggio and Rodolfo Zunino},
  title		= {{SOM}-based Interpolation for Image Compression},
  volume	= {I},
  pages		= {739--742},
  booktitle	= {Proc. WCNN'95, World Congress on Neural Networks},
  year		= {1995},
  organization	= {INNS},
  dbinsdate	= {oldtimer}
}

@InCollection{	  anouar96a,
  author	= {F. Anouar and F. Badran and S. Thiria},
  title		= {Topological maps for mixture densities},
  booktitle	= {Artificial Neural Networks---ICANN 96. 1996 International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1996},
  editor	= {C. {von der Malsburg} and W. {von Seelen} and J. C.
		  Vorbruggen and B. Sendhoff},
  address	= {Berlin, Germany},
  pages		= {833--8},
  dbinsdate	= {oldtimer}
}

@InCollection{	  anouar97a,
  author	= {F. Anouar and F. Badran and S. Thiria},
  title		= {Probabilistic self organized map. {A}pplication to
		  classification},
  booktitle	= {5th European Symposium on Artificial Neural Networks ESANN
		  '97. Proceedings},
  publisher	= {D facto},
  year		= {1997},
  editor	= {M. Verleysen},
  address	= {Brussels, Belgium},
  pages		= {13--18},
  dbinsdate	= {oldtimer}
}

@InCollection{	  anouar97b,
  author	= {F. Anouar and F. Badran and S. Thiria},
  title		= {Self organizing map, a probabilistic approach},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {339--344},
  dbinsdate	= {oldtimer}
}

@Article{	  anouar98a,
  author	= {Anouar, F. and Badran, F. and Thiria, S.},
  title		= {Probabilistic \mbox{self-organizing} map and radial basis
		  function networks},
  journal	= {Neurocomputing},
  year		= {1998},
  number	= {1},
  volume	= {20},
  pages		= {83--96},
  abstract	= {We propose in this paper a new learning algorithm
		  probabilistic self-organizing map (PRSOM) using a
		  probabilistic formalism for topological maps. This
		  algorithm approximates the density distribution of the
		  input set with a mixture of normal distributions. The
		  unsupervised learning is based on the dynamic clusters
		  principle and optimizes the likelihood function. A
		  supervised version of this algorithm based on radial basis
		  functions (RBF) is proposed. In order to validate the
		  theoretical approach, we achieve regression tasks on
		  simulated and real data using the PRSOM algorithm.
		  Moreover, our results are compared with normalized Gaussian
		  basis functions (NGBF) algorithm.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ansamaki00a,
  author	= {Ansamaki, J. and Toivanen, P. J. and Parkinnen, J.},
  title		= {New multispectral edge detection methods based on
		  roughness and differential Prewitt},
  booktitle	= {Signal Processing X Theories and Applications. Proceedings
		  of EUSIPCO 2000. Tenth European Signal Processing
		  Conference. Tampere Univ. Technology, Tampere, Finland},
  year		= {2000},
  volume	= {3},
  pages		= {1349--52},
  abstract	= {Two new edge detection methods for multispectral images
		  are presented. The first method is based on calculating
		  Euclidean distances between pixel vectors inside a suitable
		  mask. This method approximates the roughness of the pixel
		  neighborhood. The result is a scalar edge image. This
		  method is tested using an airborne remote sensing image and
		  two artificial images. One artificial image contains
		  metameric colors and the other has a group of colors
		  arranged using the self organizing map. In all the cases,
		  the edges are clearly visible without further filtering.
		  The second method is a differential Prewitt method using
		  vectors for calculating the gradient value. In this case
		  the edge image is a vector image. This method is tested
		  using the same test images as in the first case.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  ansari90a,
  author	= {N. Ansari and Y. Chen},
  title		= {A neural network model to configure maps for a satellite
		  communication network},
  booktitle	= {Proc. GLOBECOM'90, IEEE Global Telecommunications
		  Conference and Exhibition. 'Communications: Connecting the
		  Future'},
  year		= {1990},
  volume	= {II},
  pages		= {1042--1046},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ansari91a,
  author	= {N. Ansari and D. Liu},
  title		= {The performance evaluation of a new neural network based
		  traffic management scheme for a satellite communication
		  network},
  booktitle	= {Proc. GLOBECOM'91, IEEE Global Telecommunications
		  Conference Countdown to the New Millennium. Featuring a
		  Mini-Theme on: 'Personal Communications Services (PCS). '},
  year		= {1991},
  volume	= {I},
  pages		= {110--114},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  ansari95a,
  author	= {Ansari, N. and Dequan Liu},
  title		= {The performance evaluation of a new neural network-based
		  traffic management scheme for a satellite communication
		  network},
  journal	= {Neurocomputing},
  year		= {1995},
  volume	= {8},
  number	= {3},
  pages		= {263--82},
  month		= {Aug},
  dbinsdate	= {oldtimer}
}

@Article{	  antikainen00a,
  author	= {Antikainen, O. K. and Rantanen, J. T. and Yliruusi, J.
		  K.},
  title		= {Use of the Kohonen self-organising map to predict the
		  flowability of powders},
  journal	= {STP PHARMA SCIENCES},
  year		= {2000},
  volume	= {10},
  number	= {5},
  month		= {SEP-OCT},
  pages		= {349--354},
  abstract	= {A true-dimensional Kohonen self-organising map was used to
		  predict weight variation in tablets on the basis of flow
		  properties of powders. Flow test data of the powders
		  consisted of four different angles of repose, flowing time
		  and the non- uniformity of flow determined front thirty-six
		  powder batches. The network was trained with three
		  microcrystalline cellulose grades and their mixture.
		  Microcrystalline cellulose fillers contained various
		  amounts of a poor flowing drug substance, paracetamol. The
		  final network was tested using a few silicified
		  microcrystalline cellulose powders which also contained
		  paracetamol. The Kohonen self-organising map is able to
		  combine numerous parameters that describe powder
		  flowability, and to present the data in an
		  easily-understood, visual form. It is practically
		  impossible for any other technique to achieve this. The
		  self-organising map appeared to he able to cluster
		  flowability data so that the best flowing powders were
		  localised in the same part on the map. The predictive
		  ability of the map was satisfactory.},
  dbinsdate	= {2002/1}
}

@Article{	  antonini90a,
  author	= {M. Antonini and M. Barlaud and P. Mathieu and J. C.
		  Feauveau},
  title		= {Multiscale image coding using the {K}ohonen Neural
		  Network},
  journal	= {Proc. SPIE---The International Society for Optical
		  Engineering},
  year		= {1990},
  volume	= {1360},
  number	= {1},
  pages		= {14--26},
  note		= {Conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  antonini90b,
  author	= {M. Antonini and M. Barlaud and P. Mathieu},
  title		= {Predictive interscale image coding using vector
		  quantization},
  booktitle	= {Signal Processing V. Theories and Applications. Proc.
		  EUSIPCO-90, Fifth European Signal Processing Conference},
  year		= {1990},
  editor	= {L. Torres and E. Masgrau and M. A. Lagunas},
  volume	= {II},
  pages		= {1091--1094},
  organization	= {CIDEM; CIRIT; IBM; et al},
  publisher	= {Elsevier},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@Article{	  anzali98a,
  author	= {S. Anzali and W. W. K. R. Mederski and M. Osswald and D.
		  Dorsch},
  title		= {1 Endothelin Antagonists Search for Surrogates of
		  Methylendioxyphenyl by Means of a {K}ohonen Neural
		  Network},
  journal	= {Bioorganic \& Medicinal Chemistry Letters},
  volume	= {8},
  pages		= {11--16},
  year		= {1998},
  dbinsdate	= {oldtimer}
}

@Article{	  anzali98b,
  author	= {S. Anzali and J. Gasteiger and U. Holzgrabe and J.
		  Polanski and J. Sadowski and A. Teckentrup and M. Wagener},
  title		= {The Use of Self Organizing Neural Networks in Drug
		  Design},
  journal	= {Perspectives in Drug Discovery and Design},
  volume	= {9--11},
  pages		= {273--299},
  year		= {1998},
  dbinsdate	= {oldtimer}
}

@Article{	  aoki00a,
  author	= {Aoki, H. and Saito, T.},
  title		= {On classification function of self-organizing map with
		  threshold operation},
  journal	= {Transactions-of-the-Institute-of-Electronics,-Information-and-Communication-Engineers-A}
		  ,
  year		= {2000},
  volume	= {},
  pages		= {1122--4},
  abstract	= {For a basic self-organizing map, we propose a simple
		  learning algorithm with threshold operation. Preliminary
		  numerical experimental results suggest that, adjusting the
		  threshold appropriately, undesired interventions among
		  input classes can be suppressed and the classification
		  function can be reinforced.},
  dbinsdate	= {2002/1}
}

@Article{	  ara97a,
  author	= {M. Ara and N. Suzuki and E. Suzuki and H. Mukae},
  title		= {Application of \mbox{self-organizing} feature map to
		  failure diagnosis through sound data},
  journal	= {Research Reports of Kogakuin University},
  year		= {1997},
  publisher	= {Kogakuin Univ},
  volume	= {4},
  number	= {82},
  pages		= {129--33},
  dbinsdate	= {oldtimer}
}

@Article{	  arabshahi96a,
  author	= {Payman Arabshahi and Jai J. Choi and Robert J. {Marks II}
		  and Thomas P. Caudell},
  title		= {Fuzzy Parameter Adaptation in Optimization: {SOM} Neural
		  Net Training Examples},
  journal	= {IEEE Computational Science \& Engineering},
  year		= 1996,
  volume	= 3,
  pages		= {57--65},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  arai00a,
  author	= {Arai, Y. and Hakura, J.},
  title		= {Teleoperation system for real world robots-adaptive robot
		  navigation based on sensor fusion},
  booktitle	= {Proceedings Seventh International Conference on Parallel
		  and Distributed Systems: Workshops. IEEE Comput. Soc, Los
		  Alamitos, CA, USA},
  year		= {2000},
  volume	= {},
  pages		= {487--92},
  abstract	= {The authors propose a teleoperation system with an
		  autonomous robot which is able to solve tasks even without
		  a large load for the operator and the system. Most
		  teleoperation systems require skilled operators and
		  expensive interfaces to solve tasks because they assume
		  that the operator controls a robot completely. For these
		  problems, we propose a teleoperation system which consists
		  of an operation system and an autonomous robot. The
		  operation system has a man-machine interface and allows a
		  user to specify the working space and the tasks to be done.
		  The autonomous robot follows the instruction from the
		  operation system to solve the specific tasks. The paper
		  focuses on navigation problems of the autonomous robot as
		  an essential part of the proposed system. Namely, the
		  autonomous robot should keep on the instructed paths in the
		  real world to achieve a goal of the tasks. Our approach is
		  based on a sensor fusion method based on two learning
		  schemes: self-organizing map (SOM) and reinforcement
		  learning. These learning schemes allow the system to be
		  able to solve the tasks in an unreliable environment such
		  as outdoors. Computational simulations reveal the
		  effectiveness and robustness of the proposed method in the
		  navigation problem.},
  dbinsdate	= {2002/1}
}

@Article{	  arakawa01a,
  author	= {Arakawa, Y. and Yoshioka, M. and Omatu, S.},
  title		= {Land cover mapping of synthetic aperture radar data by
		  using texture analysis and self-organizing map},
  journal	= {Transactions-of-the-Institute-of-Electrical-Engineers-of-Japan,-Part-C}
		  ,
  year		= {2001},
  volume	= {121},
  pages		= {234--9},
  abstract	= {We propose a method to apply a texture analysis to land
		  cover mapping of synthetic aperture radar (SAR) data.
		  Texture is a fundamental characteristic in image analysis.
		  By using the texture analysis it is possible to find
		  patterns in an aerial image. For classification of
		  categories we adopt the Kohonen's self-organizing map and
		  compare the results with those of the maximum likelihood
		  classifier.},
  dbinsdate	= {2002/1}
}

@Article{	  araki96a,
  author	= {H. Araki and H. Fukumoto and T. Ae},
  title		= {Image processing using simplified {K}ohonen network},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1996},
  volume	= {2661},
  pages		= {24--33},
  note		= {(Real-Time Imaging Conf. Date: 29--30 Jan. 1996 Conf. Loc:
		  San Jose, CA, USA Conf. Sponsor: SPIE; Soc. Imaging Sci. \&
		  Technol)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  aras00a,
  author	= {Aras, N. and Altmel, I. K. and Oommen, J.},
  title		= {A Kohonen-like decomposition method for the traveling
		  salesman problem {KNIES}-{DECOMPOSE}},
  booktitle	= {ECAI 2000. 14th European Conference on Artificial
		  Intelligence. including Prestigious Applications of
		  Intelligent Systems (PAIS-2000). Proceedings (Frontiers in
		  Artificial Intelligence and Applications Vol.54). IOS
		  Press, Amsterdam, Netherlands},
  year		= {2000},
  volume	= {},
  pages		= {261--5},
  abstract	= {In addition to the classical heuristic algorithms of
		  operations research, there have also been several
		  approaches based on artificial neural networks which solve
		  the traveling salesman problem (TSP). Their efficiency,
		  however, decreases as the problem size (number of cities)
		  increases. An idea to reduce the complexity of a
		  large-scale TSP instance is to decompose or partition it
		  into smaller subproblems, which are easier to solve. We
		  introduce an all-neural decomposition heuristic that is
		  based on a recent self-organizing map called KNIES (Kohonen
		  Network Incorporating Explicit Statistics) (N. Aras, 1999),
		  which has been successfully implemented in solving both the
		  Euclidean TSP and the Euclidean Hamiltonian path problem.},
  dbinsdate	= {2002/1}
}

@Article{	  aras99a,
  author	= {Aras, N. and Oommen, B. J. and Altinel, I. K.},
  title		= {{K}ohonen network incorporating explicit statistics and
		  its application to the travelling salesman problem},
  journal	= {Neural Networks},
  year		= {1999},
  number	= {9},
  volume	= {12},
  pages		= {1273--1284},
  abstract	= {In this paper we introduce a new self-organizing neural
		  network, the Kohonen Network Incorporating Explicit
		  Statistics (KNIES) that is based on Kohonen's
		  Self-Organizing Map (SOM). The primary difference between
		  the {SOM} and the KNIES is the fact that every iteration in
		  the training phase includes two distinct modules---the
		  attracting module and the dispersing module. As a result of
		  the newly introduced dispersing module the neurons maintain
		  the overall statistical properties of the data points.
		  Thus, although in {SOM} the neurons individually find their
		  places both statistically and topologically, in KNIES they
		  collectively maintain their mean to be the mean of the data
		  points, which they represent. Although the scheme as it is
		  currently implemented maintains the mean as its invariant,
		  the scheme can easily be generalized to maintain higher
		  order central moments as invariants. The new scheme has
		  been used to solve the Euclidean Travelling Salesman
		  Problem (TSP). Experimental results for problems taken from
		  TSPLIB indicate that it is a very accurate NN strategy for
		  the TSP---probably the most accurate neural solutions
		  available in the literature.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  arciniegas01a,
  author	= {I. Arciniegas and B. Daniel and M. J. Embrechts},
  title		= {Exploring financial crises data with self-organisinf maps
		  ({SOM})},
  booktitle	= {Advances in Self-Organising Maps},
  pages		= {39--46},
  year		= {2001},
  editor	= {Nigel Allinson and Hujun Yin and Lesley Allinson and Jon
		  Slack},
  publisher	= {Springer},
  dbinsdate	= {2002/1}
}

@Article{	  ardizzone00a,
  author	= {Ardizzone, E. and Capra, A. and La Cascia, M.},
  title		= {Using temporal texture for content-based video retrieval},
  journal	= {Journal-of-Visual-Languages-and-Computing},
  year		= {2000},
  volume	= {11},
  pages		= {241--52},
  abstract	= {Textures evolving over time are called temporal textures
		  and are very common in everyday life. Examples are the
		  smoke flowing or the wavy water of a river. The idea
		  explored in the paper is that image features based on
		  temporal texture could allow a better performance of
		  current content based video retrieval systems that are
		  mainly based on static characteristics of representative
		  frames, like color and texture. To this aim we analyze the
		  spatio-temporal nature of texture and its application in
		  content-based access to video databases. In particular, we
		  represent temporal texture using the spatiotemporal
		  autoregressive (STAR) model and a variation of
		  self-organizing maps (SOM) where each node is an
		  autoregressive model. These representation schemes have
		  been implemented in a query by example framework to analyze
		  the weaknesses and the strengths of the different
		  approaches. Preliminary experimental results are
		  reported.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  ardizzone91a,
  author	= {E. Ardizzone and A. Chella and F. Sorbello},
  title		= {A Digital Architecture Implementing the Self-Organizing
		  Feature Maps},
  booktitle	= {Artificial Neural Networks},
  year		= {1991},
  editor	= {T. Kohonen and K. M{\"{a}}kisara and O. Simula and J.
		  Kangas},
  volume	= {I},
  pages		= {721--727},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  month		= {},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ardizzone94a,
  author	= {E. Ardizzone and A. Chella and R. Rizzo},
  title		= {Color Image Segmentation Based on a Neural Gas Network},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {II},
  pages		= {1161--1164},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, image segmentation, neural gas algorithm},
  dbinsdate	= {oldtimer}
}

@Article{	  arrigo91a,
  author	= {Arrigo, P. and Giuliano, F. and Scalia, F. and Rapallo, A.
		  and Damiani, G},
  title		= {Identification of a new motif on nucleic acid sequence
		  data using {K}ohonen's \mbox{self-organizing} map.},
  journal	= {COMP. APPLIC. BIOSCI.},
  year		= {1991},
  number	= {3},
  volume	= {7},
  pages		= {353--357},
  abstract	= {The authors present a performance test of a Kohonen
		  features map applied to the fast extraction of uncommon
		  sequences from the coding region of the human insulin
		  receptor gene. They used a network with 30 neurons and with
		  a variable input window. The program was aimed at detecting
		  unique or uncommon DNA regions present in crude sequence
		  data and was able to automatically detect the signal
		  peptide coding regions of a set of human insulin receptor
		  gene data. The testing of this program with HSIRPR cDNA
		  release (EMBL data bank) indicated the presence of unique
		  features in the signal peptide coding region. This program
		  can automatically detect "singularity" from crude
		  sequencing data and it does not require knowledge of the
		  features to be found.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  arrigo94a,
  author	= {Arrigo, P. and Giuliano, F. and Damiani, G. },
  title		= {Identification of singular domains on nucleotidic
		  sequences by {SOFM}},
  booktitle	= {IEE Colloquium on 'Molecular Bioinformatics' (Digest No.
		  1994/029)},
  year		= {1994},
  pages		= {4/1},
  organization	= {Istituto per i Circuiti Elettronici, Consiglio Nazinale
		  delle Ricerche, Genova, Italy},
  publisher	= {IEE},
  address	= {London, UK},
  abstract	= {The availability of new powerful hardware architectures
		  allows to implement efficient computational models. The
		  Artificial Neural Classifiers are one of the more
		  successful class. The peculiar characteristic of Neural
		  Classifiers induced many Computer Scientist and Molecular
		  Biologist to apply they to the analysis of many kind of
		  experimental data or to the development of integrated
		  systems of information retrieval. The Molecular Biology
		  offers an very important application field in order to test
		  the performances of different neural architectures. The
		  analysis of biosequences represents one of the more
		  important tasks. The major part of the literature on this
		  applicative field is based on the 'supervised' neural
		  models. The genomic sequences could be considered like
		  noisy signals and the knowledge about the syntactical rules
		  that regulate the gene expression are not well known. This
		  reason induced us to refuse all the knowledge about the
		  genomic features an to make a new 'blind' search on
		  complete sequence, in order to take in account the
		  contextual effect on this kind of pattern recognition. In
		  this paper we present an application of Kohonen's
		  Self-Organizing Maps (KFM) to the recognition of uncommon
		  domains on cDNA sequences.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  arrowsmith99a,
  author	= {Arrowsmith, M. J. and Varley, M. R. and Picton, P. D. and
		  Heys, J. D.},
  title		= {Hybrid neural network system for texture analysis},
  booktitle	= {Seventh International Conference on Image Processing and
		  Its Applications (Conf. Publ. No.465)},
  year		= {1999},
  publisher	= {IEE},
  address	= {London, UK},
  volume	= {1},
  pages		= {339--43},
  abstract	= {Texture classification and segmentation in digital images
		  is commonly achieved using spatial grey level dependence
		  matrices (SGLDMs), often referred to as co-occurrence
		  matrices. This involves the computation of many matrices
		  over a range of different spatial separations and
		  orientations. The approach proposed in this paper uses a
		  hybrid neural network system, consisting of a
		  self-organising map followed by a backpropagation network,
		  to restrict the number of SGLDMs that need to be computed.
		  The system is trained in two phases on images with known
		  texture content. The trained system is able to provide
		  information, in the form of pixel spacing and orientation,
		  on the texture content of unseen images. This information
		  may be used to select appropriate SGLDMs for further
		  texture classification. Experimental results are presented
		  which demonstrate the effective performance of the
		  system.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  arroyo01a,
  author	= {Arroyo, J. M. R. and Beddoes, A. J. and Allinson, N. M.},
  title		= {Condition monitoring of 11 {KV} paper insulated cables
		  using self-organising maps},
  booktitle	= {Canadian Conference on Electrical and Computer
		  Engineering},
  year		= {2001},
  editor	= {},
  volume	= {1},
  pages		= {259--264},
  organization	= {EA Technology Ltd},
  publisher	= {},
  address	= {},
  abstract	= {This paper concerns the feasibility of using
		  Self-Organising Feature Maps for the insulation assessment
		  of paper insulated cables. This class of neural networks is
		  able to isolate different clusters within the discharge
		  activity obtained throughout a degradation process.
		  However, once trained, they are incapable of identifying
		  novel states in the insulation of the sample. As a possible
		  solution of this problem, the authors present a variation
		  of the SOM based on the expansion of the trained map. With
		  this modification, SOM can be used for the condition
		  monitoring of the cables and the prediction of incipient
		  faults.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  arroyo01b,
  author	= {Arroyo, J. M. R. and Beddoes, A. J. and Alinson, N. M.},
  title		= {Insulation condition assessment of 11 kV paper cables
		  using neural networks},
  booktitle	= {IEE Symposium Pulsed Power 2001 (Digest No.01/156). IEE,
		  London, UK},
  year		= {2001},
  volume	= {1},
  pages		= {},
  abstract	= {The increasing demand of energy and the cost associated
		  with the installation of new power systems are intensifying
		  the design and implementation of condition monitoring
		  techniques, which are able to estimate the remaining active
		  life of existing systems and, therefore, provide decision
		  support for their maintenance or replacement. For paper
		  insulated power cables, a common method for their condition
		  assessment is the detection and analysis of partial
		  discharges (PD). These discharges in the cavities of the
		  paper insulation increase the size of the voids and cause
		  the development of carbonaceous paths between the fibres of
		  the paper layers. These defects may lead to fatal breakdown
		  of the cable. Although this physical deterioration process
		  is well known, the essential parameters required for the
		  control of the different stages of the degradation remain
		  uncertain. Therefore, a large number of variables have to
		  be monitored to develop an efficient and reliable condition
		  monitoring system for paper insulated cables.
		  Self-organising maps (SOM), a type of artificial neural
		  network, are a suitable technique for the processing and
		  analysis of high dimensional problems involving large
		  amounts of data, being particularly appropriate for the
		  recognition of trends and feature clustering. They form the
		  basis of the condition monitoring technique presented in
		  this paper.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  arseneau94a,
  author	= {J. Brant Arseneau and Tim Spracklen},
  title		= {Reengineering Software Modularity using Artificicial
		  Neural Networks},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {II},
  pages		= {1384--1387},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, visualization, clustering},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  arseneau94b,
  author	= {J. Brant Arseneau and Tim Spracklen},
  title		= {Reengineering Software Modularity using Artificial Neural
		  Networks},
  booktitle	= {Proc. WCNN'94, World Congress on Neural Networks},
  year		= {1994},
  volume	= {I},
  pages		= {467--470},
  organization	= {INNS},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  annote	= {application, symbolic data, clustering},
  dbinsdate	= {oldtimer}
}

@InCollection{	  asanovic97a,
  author	= {K. Asanovic},
  title		= {A fast {K}ohonen net implementation for Spert-II},
  booktitle	= {Biological and Artificial Computation: From Neuroscience
		  to Technology. International Work Conference on Artificial
		  and Natural Neural Networks, IWANN'97. Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {J. Mira and R. Moreno-Diaz and J. Cabestany},
  address	= {Berlin, Germany},
  pages		= {792--800},
  dbinsdate	= {oldtimer}
}

@Article{	  assadollahi01a,
  author	= {Assadollahi, R. and Pulvermuller, F.},
  title		= {Neural network classification of word evoked neuromagnetic
		  brain activity},
  journal	= {Emergent neural computational architectures based on
		  neuroscience. Towards neuroscience-inspired computing.
		  Springer-Verlag, Berlin, Germany; 2001; x+576
		  pp.p.311--19},
  year		= {2001},
  volume	= {},
  pages		= {311--19},
  abstract	= {The brain-physiological signatures of words are modulated
		  by their psycholinguistic and physical properties. The
		  fine-grained differences in complex spatio-temporal
		  patterns of a single word induced brain response may be
		  detected using unsupervised neuronal networks. Objective of
		  this study was to motivate and explore an architecture of a
		  Kohonen net and its performance, even when physical
		  stimulus properties are kept constant over the classes. We
		  investigated 16 words from four lexico-semantic classes.
		  The items from the four classes were matched for word
		  length and frequency. A Kohonen net was trained on the data
		  recorded from a single subject. After learning, the network
		  performed above chance on new testing data. The results
		  obtained suggest that the research on single trial
		  recognition of brain responses is feasible and a rich field
		  to explore.},
  dbinsdate	= {2002/1}
}

@InCollection{	  atlas95a,
  author	= {L. Atlas and L. Owsley and J. McLaughlin and G. Bernard},
  title		= {Automatic feature-finding for time-frequency
		  distributions},
  booktitle	= {Proceedings of the IEEE-SP International Symposium on
		  Time-Frequency and Time-Scale Analysis},
  publisher	= {Gordon \& Breach},
  year		= {1995},
  editor	= {G. F. Forsyth and M. Ali},
  address	= {Newark, NJ, USA},
  pages		= {333--6},
  dbinsdate	= {oldtimer}
}

@InCollection{	  atmaca96a,
  author	= {H. Atmaca and M. Bulut and D. Demir and S. Pazar},
  title		= {A new fuzzy {K}ohonen clustering network based on
		  histogram for image segmentation},
  booktitle	= {Proceedings of the Eleventh International Symposium on
		  Computer and Information Sciences. ISCIS},
  publisher	= {Middle East Tech. Univ},
  year		= {1996},
  volume	= {2},
  editor	= {V. Atalay and U. Halici and K. Inan and N. Yalabik and A.
		  Yazici},
  address	= {Ankara, Turkey},
  pages		= {845--9},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  atsalakis01a,
  author	= {Atsalakis, A. and Andreadis, I. and Papamarkos, N.},
  title		= {Histogram based color reduction through Self-Organized
		  neural networks},
  booktitle	= {ARTIFICAL NEURAL NETWORKS-ICANN 2001, PROCEEDINGS},
  year		= {2001},
  pages		= {470--476},
  abstract	= {A now technique suitable for reduction of the number of
		  colors in an image is presented in this paper. It is based
		  on histogram processing and the use of Kohonen Self
		  Organizing Feature Map (SOFM) neural networks. Initially,
		  the dominant colors of each primary image are extracted
		  through a simple linear piece-wise histogram approximation
		  process. Then, using a SOFM the dominant color components
		  of each primary color band are obtained and a look up table
		  is constructed containing all possible color triplets. The
		  final dominant colors are extracted from the look-up table
		  entries using a SOFM by specifying the number of output
		  neurons equal to the number of the dominant colors. Thus,
		  the final image has all the dominant color classes,
		  Experimental and comparative results demonstrate the
		  applicability of the proposed technique.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  atsalakis01b,
  author	= {Atsalakis, A. and Papamarkos, N. and Strouthopoulos, C.},
  title		= {A new adaptive color quantization technique},
  booktitle	= {ARTIFICAL NEURAL NETWORKS-ICANN 2001, PROCEEDINGS},
  year		= {2001},
  pages		= {1006--1012},
  abstract	= {This paper proposes a new algorithm for color
		  quantization. The proposed approach achieves color
		  quantization using an adaptive tree clustering procedure.
		  In each node of the tree a self- organized Neural Network
		  Classifier (NNC) is used which is fed by image color values
		  and additional local spatial features. The NNC consists of
		  a Principal Component Analyzer (PCA) and a Kohonen
		  self-organized feature map (SOFM) neural network. The
		  output neurons of the NNTC define the color classes for
		  each node. The final image not only has the dominant image
		  colors, but also its texture approaches the image local
		  characteristics used. For better classification, split and
		  merging conditions are used in order to define if color
		  classes must be split or merged. To speed-up the entire
		  algorithm and reduce memory requirements, a fractal
		  scanning sub-sampling technique is used.},
  dbinsdate	= {2002/1}
}

@Article{	  audouze00a,
  author	= {Audouze, K. and Ros, F. and Pintore, M. and Chretien,
		  J.R.},
  title		= {Prediction of odours of aliphatic alcohols and
		  carbonylated compounds using fuzzy partition and self
		  organising maps ({SOM})},
  journal	= {Analusis},
  year		= {2000},
  volume	= {28},
  number	= {7},
  month		= {Sep},
  pages		= {625--632},
  organization	= {Univ of Orleans},
  publisher	= {EDP Sci},
  address	= {New York, NY},
  abstract	= {A set of 114 olfactory molecules divided into fruity,
		  ethereal and camphoraceous compounds, was submitted to an
		  analysis by Kohonen Neural Networks, also known as Self
		  Organising Map (SOM). The compounds are represented in a
		  hyperspace derived from their molecular descriptors and SOM
		  gives a useful projection of this hyperspace onto a 2D map.
		  Owing to the complexity of the olfaction mechanism,
		  evidenced by the fact that one compound can exhibit
		  simultaneously different properties, SOM alone is unable to
		  take into account the olfaction diversity of the original
		  114 compounds. Then, a Fuzzy Partition method was applied
		  on the Kohonen map previously developed. The obtained
		  results allowed delineating different representative zones
		  for the three odours, expressing more closely the olfactory
		  richness. The ability of the Hybrid System combining SOM
		  and Fuzzy Partition to model the three odours was validated
		  by dividing the 114 compounds into a training set and a
		  test set, including 86 and 28 molecules, respectively. The
		  most important olfactory characteristics were reproduced
		  satisfactorily for the entire test set compounds.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  auger91a,
  author	= {J. M. Auger},
  title		= {Parallel implementation on transputer of {K}ohonen's
		  algorithm},
  booktitle	= {Computing with Parallel Architectures: T. Node},
  year		= {1991},
  editor	= {D. Gassilloud and J. C. Grossetie},
  pages		= {215--226},
  organization	= {Inst. Syst. Eng. Inf},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  auger92a,
  author	= {Jean-Marie Auger and Yizhak Idan and Raymond Chevallier
		  and Bernadette Dorizzi},
  title		= {Complementary Aspects of Topological Maps and Time Delay
		  Neural Networks for Character Recognition},
  booktitle	= {Proc. IJCNN'92, International Joint Conference on Neural
		  Networks},
  year		= {1992},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  volume	= {IV},
  pages		= {444--449},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  augusteijn93a,
  author	= {Marijke F. Augusteijn and Tammy L. Skufca},
  title		= {Identification of Human Faces through Texture-Based
		  Feature Recognition and Neural Network Technology},
  booktitle	= {Proc. ICNN'93, International Conference on Neural
		  Networks},
  year		= {1993},
  volume	= {I},
  pages		= {392--398},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InCollection{	  augusteijn95a,
  author	= {M. F. Augusteijn and K. K. McCarthy},
  title		= {Image indexing applied to character font recognition by
		  means of a {K}ohonen neural network hierarchy},
  booktitle	= {Intelligent Engineering Systems Through Artificial Neural
		  Networks. Vol. 5. Fuzzy Logic and Evolutionary Programming.
		  Proceedings of the Artificial Neural Networks in
		  Engineering (ANNIE'95)},
  publisher	= {ASME Press},
  year		= {1995},
  editor	= {C. H. Dagli and M. Akay and C. L. P. Chen and B. R.
		  Fernandez and J. Ghosh},
  address	= {New York, NY, USA},
  pages		= {431--6},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  aupetit00a,
  author	= {Michael Aupetit and Pierre Couturier and Pierre Massotte},
  title		= {Function Approximation with Continuous Self-Organizing
		  Maps using Neighboring Influence Interpolation},
  booktitle	= {Proceeding of the ICSC Symposia on Neural Computation
		  (NC'2000) May 23-26, 2000 in Berlin, Germany},
  editor	= {H. Bothe and R. Rojas},
  year		= {2000},
  organization	= {LGI2P - Site EERIE - EMA},
  publisher	= {ICSC Academic Press},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  aupetit01a,
  author	= {M. Aupetit and P. Couturier and P. Massotte},
  title		= {Vector quantisation with $\gamma$-observable neighbors},
  booktitle	= {Advances in Self-Organising Maps},
  pages		= {230--7},
  year		= {2001},
  editor	= {Nigel Allison and Hujun Yin and Lesley Allison and Jon
		  Slack},
  publisher	= {Springer},
  dbinsdate	= {2002/1}
}

@InCollection{	  austermeier96a,
  author	= {H. Austermeier and G. Hartmann and R. Hilker},
  title		= {Color-calibration of a robot vision system using
		  \mbox{self-organizing} feature maps},
  booktitle	= {Artificial Neural Networks---ICANN 96. 1996 International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1996},
  editor	= {C. {von der Malsburg} and W. {von Seelen} and J. C.
		  Vorbruggen and B. Sendhoff},
  address	= {Berlin, Germany},
  pages		= {257--62},
  dbinsdate	= {oldtimer}
}

@TechReport{	  autere90a,
  author	= {Antti Autere and Jarmo T. Alander and Lasse
		  Holmstr{\"{o}}m and Peter Holmstr{\"{o}}m and Ari
		  H{\"{a}}m{\"{a}}l{\"{a}}inen and Juha Tuominen},
  title		= {Surface Type Recognition by a Hair Sensor},
  type		= {Res. Reports},
  number	= {A2},
  institution	= {Rolf Nevanlinna Institute},
  address	= {Helsinki, Finland},
  year		= {1990},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  avcibas00a,
  author	= {Avcibas, I. and Sankur, B.},
  title		= {Statistical analysis of image quality measures},
  booktitle	= {Signal Processing X Theories and Applications. Proceedings
		  of EUSIPCO 2000. Tenth European Signal Processing
		  Conference. Tampere Univ. Technology, Tampere, Finland},
  year		= {2000},
  volume	= {4},
  pages		= {2181--4},
  abstract	= {In this paper, we conduct a statistical analysis of the
		  sensitivity and consistency behavior of objective image
		  quality measures. We categorize the quality measures and
		  compare them for still image compression applications. The
		  measures have been categorized into pixel difference-based,
		  correlation-based, edge-based, spectral-based,
		  context-based and HVS-based (human visual system-based)
		  measures. The mutual relationships between the measures are
		  visualized by plotting their Kohonen maps. Their
		  consistency and sensitivity to coding as well as additive
		  noise and blur are investigated via ANOVA analysis of their
		  scores. It has been found that measures based on HVS, on
		  phase spectrum and on multiresolution mean square error are
		  most discriminative to coding artifacts.},
  dbinsdate	= {2002/1}
}

@Article{	  avcibas02a,
  author	= {Avcibas, I. and Sankur, B. and Sayood, K.},
  title		= {Statistical evaluation of image quality measures},
  journal	= {Journal-of-Electronic-Imaging},
  year		= {2002},
  volume	= {11},
  pages		= {206--23},
  abstract	= {In this work we comprehensively categorize image quality
		  measures, extend measures defined for gray scale images to
		  their multispectral case, and propose novel image quality
		  measures. They are categorized into pixel difference-based,
		  correlation-based, edge-based, spectral-based,
		  context-based and human visual system (HVS)-based measures.
		  Furthermore we compare these measures statistically for
		  still image compression applications. The statistical
		  behavior of the measures and their sensitivity to coding
		  artifacts are investigated via analysis of variance
		  techniques. Their similarities or differences are
		  illustrated by plotting their Kohonen maps. Measures that
		  give consistent scores across an image class and that are
		  sensitive to coding artifacts are pointed out. It was found
		  that measures based on the phase spectrum, the
		  multiresolution distance or the HVS filtered mean square
		  error are computationally simple and are more responsive to
		  coding artifacts. We also demonstrate the utility of
		  combining selected quality metrics in building a
		  steganalysis tool.},
  dbinsdate	= {2002/1}
}

@InCollection{	  azam98a,
  author	= {F. Azam and H. F. Van Landingham},
  title		= {Adaptive self organizing feature map neuro-fuzzy technique
		  for dynamic system identification},
  booktitle	= {Proceedings of the 1998 IEEE International Symposium on
		  Intelligent Control (ISIC) held jointly with IEEE
		  International Symposium on Computational Intelligence in
		  Robotics and Automation (CIRA) Intelligent Systems and
		  Semiotics (ISAS)},
  publisher	= {IEEE},
  year		= {1998},
  address	= {New York, NY, USA},
  pages		= {337--41},
  dbinsdate	= {oldtimer}
}

@InCollection{	  azam98b,
  author	= {Farooq Azam and H. F. VanLandingham},
  title		= {A New Approach for Modular Neural Network Design and
		  Learning},
  booktitle	= {Proc. JCIS'98},
  publisher	= {Association for Intelligent Machinery, Inc},
  year		= 1998,
  editor	= {Paul P. Wang},
  volume	= {II},
  pages		= {147--189},
  abstract	= {This paper presents a new efficient design and learning
		  methodology for a modularly structured neural network. This
		  design scheme is based on the input space partitioning
		  using an adaptive fuzzy partitioning module implementing an
		  adaptive self organizing feature map paradigm. The input
		  space partitioning serves the dual purpose of the
		  determination of optimal number of expert neural networks
		  in the modular structure along with facilitation of the
		  neural architecture training. This approach leads to a
		  simple and optimal modular neural architecture along with a
		  simplified training algorithm as compared to generally used
		  expectation maximization (EM) algorithm. The experimental
		  results indicate that this approach has good capabilities
		  for nonlinear system identification.},
  dbinsdate	= {oldtimer}
}

@Article{	  azami00a,
  author	= {Azami, Seyed Bahram Zahir and Feng, Gang},
  title		= {Robust vector quantizer design using self-organizing
		  neural networks},
  journal	= {Signal Processing},
  year		= {2000},
  volume	= {80},
  number	= {7},
  month		= {Jul},
  pages		= {1289--1298},
  organization	= {Univ of Ottawa},
  publisher	= {Elsevier Science Publishers B.V.},
  address	= {Amsterdam},
  abstract	= {In this paper we propose a new method to design vector
		  quantizers for noisy channels. Self-organizing neural
		  networks are known for their efficiency in voice and image
		  data compression; we use self-organizing algorithm to
		  create a topological similarity between the input space and
		  the index space. This similarity reduces the effect of
		  channel noise because any single bit error in a transmitted
		  index will be translated to a close codevector in the input
		  space which yields relatively small distortion. For an
		  8-bit vector quantizer, the proposed system resulted in
		  4.59 dB spectral distortion in a highly noisy channel while
		  a simple LBG, LBG with splitting and 2-D self-organizing
		  map (SOM) resulted in 5.96, 5.46 and 5.02 dB of distortion,
		  respectively.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  azcarraga00a,
  author	= {Azcarraga, Arnulfo P.},
  title		= {Assessing self-organization using order metrics},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2000},
  editor	= {},
  volume	= {6},
  pages		= {159--164},
  organization	= {Natl Univ of Singapore},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {Self-Organizing Maps (SOM) are proving to be useful as
		  data analysis and visualization tools because they can
		  visually render the data analysis results in 2D or 3D, and
		  do not need category information for each input pattern.
		  But this unsupervised nature of the training process makes
		  it difficult to have separate training and test sets to
		  determine the quality of the training process, which is
		  done quite naturally for supervised Neural Network learning
		  algorithms. In applications like data analysis, where there
		  is little clue as to the way the SOM is supposed to look
		  like after training, it is important to be able to assess
		  the quality of the self-organization process independent of
		  the application, and without need for category information.
		  The Average Unit Disorder has been used to assess the
		  quality of the ordering of a self-organized map. It is
		  shown here that this same order metric can be used to
		  assess the quality of the self-organization process itself.
		  Based on this order metric, it can be determined whether
		  the SOM has adequately learned, whether the parameters used
		  to train the SOM have been correctly specified, and whether
		  the SOM variant used is well-suited to the specific problem
		  at hand.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  azcarraga00b,
  author	= {Azcarraga, A. P.},
  title		= {Evaluating {SOM}s using order metrics},
  booktitle	= {8th European Symposium on Artificial Neural Networks.
		  ESANN"2000. Proceedings. D-Facto, Brussels, Belgium},
  year		= {2000},
  volume	= {},
  pages		= {255--60},
  abstract	= {It has been shown that self-organized maps, when
		  adequately trained with the set of integers 1 to 32, lay
		  out real numbers in a 2D map in an ordering that is
		  superior to any of the known 2D orderings, such as the
		  Cantor-diagonal, Morton, Peano-Hilbert, raster-scan,
		  row-prime, spiral, and random orderings. Two 2D order
		  metrics (average direct neighbor distance and average unit
		  disorder) have been used to assess the quality of a map's
		  2D ordering. It is shown that these same order metrics are
		  useful in assessing the quality of the self-organization
		  process itself. Based on these metrics, it can be
		  determined whether the SOM has already adequately learned
		  and whether the parameters used to train the SOM have been
		  correctly specified. In applications like data analysis,
		  where there is little clue as to the way the SOM is
		  supposed to look like after training, it is important to be
		  able to assess the quality of the self-organization process
		  independent of the application.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  azcarraga01a,
  author	= {Azcarraga, A. P. and Yap, T. N. Jr},
  title		= {Comparing keyword extraction techniques for {WEBSOM} text
		  archives},
  booktitle	= {Proceedings 13th IEEE International Conference on Tools
		  with Artificial Intelligence. ICTAI 2001. IEEE Comput. Soc,
		  Los Alamitos, CA, USA},
  year		= {2001},
  volume	= {},
  pages		= {187--94},
  abstract	= {The WEBSOM methodology for building very large text
		  archives has a very slow method for extracting meaningful
		  unit labels. This is because the method computes for the
		  relative frequencies of all the words of all the documents
		  associated to each unit and then compares these to the
		  relative frequencies of all the words of all the other
		  units of the map. Since maps may have more than 100,000
		  units and the archive may contain up to 7 million
		  documents, the existing WEBSOM method is not practical. A
		  fast alternative method is based on the distribution of
		  weights in the weight vectors of the trained map, plus a
		  simple manipulation of the random projection matrix used
		  for input data compression. Comparisons made using a WEBSOM
		  archive of the Reuters text collection reveal that a high
		  percentage of keywords extracted using this method match
		  the keywords extracted for the same map units using the
		  original WEBSOM method.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  azcarraga01b,
  author	= {Azcarraga, A. P. and Yap, T. N. Jr},
  title		= {{SOM}-based methodology for building large text archives},
  booktitle	= {Proceedings Seventh International Conference on Database
		  Systems for Advanced Applications. DASFAA 2001. IEEE
		  Comput. Soc, Los Alamitos, CA, USA},
  year		= {2001},
  volume	= {},
  pages		= {66--73},
  abstract	= {Not only have self-organizing maps (SOMs), such as the
		  WEBSOM, been shown to scale up to very large datasets,
		  these maps also allow for a novel mode of navigating
		  through a large collection of text documents. The entire
		  text collection is presented to a user as a regular map,
		  where each point in the map is associated to a group of
		  documents that are likely to be composed of similar terms
		  and phrases. In addition, the closer two points are in the
		  map, the more similar are their respective associated
		  documents. Thus, once an interesting document is found in
		  the map, the user just has to click around the vicinity of
		  that document to retrieve other similar documents. A major
		  drawback of SOMs, however, is the long training time
		  required, especially for document collections where both
		  the volume and the dimensionality are huge. We demonstrate
		  how the size of the initial text collection is
		  progressively and drastically reduced from the raw document
		  collection to the final SOM-based text archive. We
		  demonstrate this using a widely studied Reuters
		  collection.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  azcarraga90a,
  author	= {A. P. Azcarraga and B. Amy},
  title		= {{K}ohonen Features Maps: toward invariant character
		  recognition},
  booktitle	= {Artificial Intelligence IV. Methodology, Systems,
		  Applications. Proc. of the Fourth International Conference
		  (AIMSA '90)},
  year		= {1990},
  editor	= {P. Jorrand and V. Sgurev},
  pages		= {209--217},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  azema-barac92a,
  author	= {M. E. Azema-Barac},
  title		= {A conceptual framework for implementing neural networks on
		  massively parallel machines},
  booktitle	= {Proc. Sixth Int. Parallel Processing Symp. },
  year		= {1992},
  editor	= {V. K. Prasanna and L. H. Canter},
  pages		= {527--530},
  publisher	= {IEEE Computer Soc. Press},
  address	= {Los Alamitos, CA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  azema-barac92b,
  author	= {M. E. Azema-Barac},
  title		= {A generic strategy for mapping neural network models on
		  transputer-based machines},
  booktitle	= {Transputing in numerical and neural network applications},
  year		= {1992},
  pages		= {244--9},
  editor	= {Reijns, G. L. and Jian Luo},
  publisher	= {IOS Press},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@InCollection{	  azimi-sadjadi96a,
  author	= {M. R. Azimi-Sadjadi and M. A. Shaikh and Bin Tian and K.
		  E. Eis and D. Reinke},
  title		= {Neural network-based cloud detection/classification using
		  textural and spectral features},
  booktitle	= {IGARSS '96. 1996 International Geoscience and Remote
		  Sensing Symposium. Remote Sensing for a Sustainable
		  Future},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {2},
  editor	= {T. I. Stein},
  address	= {New York, NY, USA},
  pages		= {1105--7},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baader94a,
  author	= {Baader, A. and Hirzinger, G. },
  title		= {A \mbox{self-organizing} algorithm for multisensory
		  surface reconstruction},
  booktitle	= {IROS '94. Proceedings of the IEEE/RSJ/GI International
		  Conference on Intelligent Robots and Systems. Advanced
		  Robotic Systems and the Real World},
  year		= {1994},
  volume	= {1},
  pages		= {81--8},
  organization	= {Inst. for Roboptics \& Syst. Dynamics, German Aerosp. Res.
		  Establ. , Wessling, Germany},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baader95a,
  author	= {Baader, A. and Hirzinger, G. },
  title		= {World modeling for a sensor-in-hand robot arm},
  booktitle	= {Proceedings of the 1995 IEEE/RSJ International Conference
		  on Intelligent Robots and Systems. Human Robot Interaction
		  and Cooperative Robots},
  year		= {1995},
  volume	= {2},
  pages		= {110--15},
  organization	= {Inst. of Robotics \& Syst. Dynamics, German Aerosp. Res.
		  Establ. , Wessling, Germany},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  babri96a,
  author	= {H. A. Babri and A. A. Osman-Gani},
  title		= {Decision making using neural networks: an application to
		  cross-cultural management},
  booktitle	= {ICNN 96. The 1996 IEEE International Conference on Neural
		  Networks},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {4},
  address	= {New York, NY, USA},
  pages		= {2060--5},
  abstract	= {Clustering various countries according to their relative
		  similarity in terms of relevant organizational variables is
		  a very useful management tool for multinational
		  enterprises(MNs). The effects of the nature of population
		  and type of 'similarity' variables on the cluster
		  compositions are generally well understood. However, the
		  differences on cluster compositions arising from the
		  underlying differences of various techniques have not been
		  well investigated. This paper is the first empirical study
		  using neural networks (specifically Kohonen's SOFM) as a
		  tool to identify country clusters based on managers'
		  perceptions of various management and human resource
		  development(HRD) practices in a large MNE. A method of
		  obtaining near-optimum number of country clusters is
		  described. The clusters developed by the SOFM network are
		  also compared with those obtained using a popular
		  clustering technique such as Q-factor analysis.},
  dbinsdate	= {oldtimer}
}

@Article{	  babu97a,
  author	= {G. P. Babu},
  title		= {Self-organizing neural networks for spatial data},
  journal	= {Pattern Recognition Letters},
  year		= {1997},
  volume	= {18},
  number	= {2},
  pages		= {133--42},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bachmann01a,
  author	= {Bachmann, Charles M. and Fusina, Robert A. and Donato,
		  Timothy F.},
  title		= {Effects of time series imagery on automated classification
		  of coastal wetland environments using projection pursuit
		  methods},
  booktitle	= {International Geoscience and Remote Sensing Symposium
		  (IGARSS)},
  year		= {2001},
  editor	= {},
  volume	= {4},
  pages		= {1868--1870},
  organization	= {Naval Research Laboratory, Remote Sensing Division, Code
		  7212},
  publisher	= {Institute of Electrical and Electronics Engineers Inc.},
  address	= {},
  abstract	= {Using a time series of Radarsat SAR and Land-sat TM
		  imagery, we developed automated classification models for
		  coastal land-cover in Northampton County, VA. The database
		  consisted of three pairs of images from each sensor taken
		  within one week of each other. We compared the results of
		  models based on data from one set with the results based on
		  data from all three sets for Radarsat only, Landsat only
		  and both sensors combined. For Radarsat only, there is an
		  increase on the order of 23--30% in the overall
		  classification score for the three-date set compared to the
		  one-date set. This is consistent with previous results
		  based on different classification methods. We found a
		  smaller increase, on the order of 1--5%, for the Landsat
		  only. For combined sensor cases, three-date Inputs did not
		  improve results when compared with single-date inputs. For
		  the feature extraction stage, we compared a Projection
		  Pursuit method with Principal Component Analysis. In the
		  final classifier stage, we compared a vector quantization
		  algorithm, LVQ, and the backward propagation of error model
		  with a cross-entropy cost function. The differences given
		  above for one-date versus three-date results hold for all
		  the methods studied. Best model agreement with a National
		  Wetlands Inventory (NWI) map was found for three-date
		  atmospherically corrected Landsat inputs, for which
		  generalization performance was 80.5% of pixels correct,
		  with 86.4% obtained on the cross-validation set used to
		  find a stopping point for model optimization, and 87.9% for
		  the training set.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  back94a,
  author	= {B. Back and G. Oosterom and K. Sere and M. {van Wezel}},
  title		= {A Comparative Study of Neural Networks in Bankrupty
		  Prediction},
  editor	= {Christer Carlsson and Timo J{\"{a}}rvi and Tapio Reponen},
  number	= {12},
  series	= {Conference Proc. of Finnish Artificial Intelligence
		  Society},
  pages		= {140--148},
  booktitle	= {Proc. Conference on Artificial Intelligence Res. in
		  Finland},
  year		= {1994},
  publisher	= {Finnish Artificial Intelligence Society},
  address	= {Helsinki, Finland},
  annote	= {application, prediction, comparison},
  dbinsdate	= {oldtimer}
}

@InCollection{	  back96a,
  author	= {B. Back and K. Sere and H. Vanharanta},
  title		= {Data mining accounting numbers using
		  \mbox{self-organizing} maps},
  booktitle	= {STeP '96---Genes, Nets and Symbols. Finnish Artificial
		  Intelligence Conference},
  publisher	= {Univ. Vaasa},
  year		= {1996},
  editor	= {J. Alander and T. Honkela and M. Jakobsson},
  address	= {Vaasa, Finland},
  pages		= {35--47},
  dbinsdate	= {oldtimer}
}

@InCollection{	  back97a,
  author	= {Barbro Back and Kaisa Sere and Hannu Vanharanta},
  title		= {Analyzing financial performance with
		  \mbox{self-organizing} maps},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {356--361},
  dbinsdate	= {oldtimer}
}

@Article{	  back98,
  author	= {Back, B.~ and Sere, K.~ and Vanharanta, H.~},
  title		= {Managing Complexity in Large Databases Using
		  Self-Organized Maps},
  journal	= {Accounting, Management and Information Technologies},
  year		= {1998},
  volume	= {8},
  number	= {4},
  pages		= {191--210},
  dbinsdate	= {oldtimer}
}

@InCollection{	  back98a,
  author	= {B. Back and K. Sere and H. Vanharanta},
  title		= {Analyzing financial performance with
		  \mbox{self-organizing} maps},
  booktitle	= {1998 IEEE International Joint Conference on Neural
		  Networks Proceedings. IEEE World Congress on Computational
		  Intelligence},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {1},
  address	= {New York, NY, USA},
  pages		= {266--70},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  badran92a,
  author	= {Badran, F. and Thiria, S. and Main, B. },
  title		= {Smoothing by use of \mbox{self-organizing} maps},
  booktitle	= {Fifth International Conference. Neural Networks and their
		  Applications. NEURO NIMES 92},
  year		= {1992},
  pages		= {107--15},
  organization	= {CEDRIC/CNAM, Paris, France},
  publisher	= {EC2},
  address	= {Nanterre, France},
  dbinsdate	= {oldtimer}
}

@Article{	  bahlmann99a,
  author	= {Bahlmann, Claus and Heidemann, Gunther and Ritter, Helge},
  title		= {Artificial neural networks for automated quality control
		  of textile seams},
  journal	= {Pattern Recognition},
  year		= {1999},
  number	= {6},
  volume	= {32},
  pages		= {1049--1060},
  abstract	= {We present a method for an automated quality control of
		  textile seams, which is aimed to establish a standardized
		  quality measure and to lower costs in manufacturing. The
		  system consists of a suitable image acquisition setup, an
		  algorithm for locating the seam, a feature extraction stage
		  and a neural network of the self-organizing map type for
		  feature classification. A procedure to select an optimized
		  feature set carrying the information relevant for
		  classification is described.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bai01a,
  author	= {Bai Bo Feng and Guo Lie Jin and Chen Xue Jun},
  title		= {The technology and theory of online recognition of
		  gas-liquid two-phase flow regime},
  booktitle	= {Proceedings-of-the-CSEE. vol.21, no.7},
  year		= {2001},
  volume	= {21},
  pages		= {46--50},
  abstract	= {It is of important industrial background and scientific
		  significance to study the online recognition of a
		  gas-liquid two-phase flow regime. The key problem of the
		  online recognition is how to map to the space of flow
		  regimes from the space of character of the parameter
		  fluctuations with the shortest time. A counter propagation
		  network (CPN) can train the results of self-organized
		  mapping with supervision. At the same time, it takes
		  advantage of a simple algorithm and doesn't require lots of
		  training samples. Therefore, it meets the online and
		  automatic recognition of gas-liquid two-phase flow regimes.
		  In the present test case, the online criterion was 8.2 s
		  with the FFT of pressure fluctuations from the vertical
		  upward section of the U-type tube. The presented technology
		  and theory has some obvious advantages, which provides a
		  feasible solution of online recognition without a databank
		  of flow regime and can partially overcome the problems
		  induced by not enough experiments in advance. In addition,
		  the problems to shorten the identification time and to
		  increase the identification possibility are discussed in
		  the paper.},
  dbinsdate	= {2002/1}
}

@InCollection{	  bailey95a,
  author	= {M. Bailey and C. Solomon and N. Kasabov and S. Greig},
  title		= {Hybrid systems for medical data analysis and decision
		  making-a case study on varicose vein disorders},
  booktitle	= {Proceedings of the Second New Zealand International
		  Two-Stream Conference on Artificial Neural Networks and
		  Expert Systems},
  publisher	= {IEEE Computer Society Press},
  year		= {1995},
  editor	= {N. K. Kasabov and G. Coghill},
  address	= {Los Alamitos, CA, USA},
  pages		= {265--8},
  dbinsdate	= {oldtimer}
}

@InCollection{	  balachander97a,
  author	= {T. Balachander and R. Kothar and H. Cualing},
  title		= {An empirical comparison of dimensionality reduction
		  techniques for pattern classification},
  booktitle	= {Artificial Neural Networks---ICANN '97. 7th International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {W. Gerstner and A. Germond and M. Hasler and J. -D.
		  Nicoud},
  address	= {Berlin, Germany},
  pages		= {589--94},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball90a,
  author	= {N. R. Ball and K. Warwick},
  title		= {Application of augmented-output self organizing feature
		  maps to the adaptive control problem},
  booktitle	= {Proc. INNC'90, Int. Neural Network Conference},
  year		= {1990},
  volume	= {I},
  pages		= {242},
  organization	= {Thomsom; SUN; British Computer Society ; et al},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball92a,
  author	= {Nigel Ball and Kevin Warwick},
  title		= {Applying \mbox{self-organizing} feature maps to the
		  control of artificial organisms in maze running tasks},
  booktitle	= {Proc. American Control Conference },
  year		= {1992},
  pages		= {3062--3063},
  organization	= {American Automatic Control Council},
  publisher	= {American Automatic Control Council},
  address	= {Green Valley, AZ},
  x		= {cp93---DIALOG No: 03560102 Variations on the now standard
		  Kobonen feature map enable an ordering of the map state
		  space by using only a limited subset of the complete input
		  vector. Also it is possible to employ merely a local
		  adaptation procedure to order the map, rather than having
		  to rely on global variables and objectives. Such variations
		  have been included as part of a Hybrid Learning System
		  (HLS) which has arisen out of a genetic-based classifier
		  system. Inthis paper a description of the modified feature
		  map is given, thisconstituting the HLS's long term memory,
		  and results on the control of simplemaze running task are
		  presented, thereby demonstrating the value of goalrelated
		  feedback within the overall network. },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball92b,
  author	= {N. R. Ball},
  title		= {Competitive Learning in Classifier Feature Maps},
  booktitle	= {Artificial Neural Networks, 2},
  year		= {1992},
  editor	= {I. Aleksander and J. Taylor},
  volume	= {I},
  pages		= {703--706},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  month		= {},
  dbinsdate	= {oldtimer}
}

@Article{	  ball92c,
  author	= {N. Ball and L. Kierman and K. Warwick and E. Cahill and D.
		  Esp and J. Macqueen},
  title		= {Neural networks for power systems alarm handling},
  journal	= {Neurocomputing},
  year		= {1992},
  volume	= {4},
  number	= {1--2},
  pages		= {5--8},
  dbinsdate	= {oldtimer}
}

@Article{	  ball93a,
  author	= {N. R. Ball and K. Warwick},
  title		= {Using \mbox{self-organizing} feature maps for the control
		  of artificial organisms},
  journal	= {IEE Proc. D (Control Theory and Applications)},
  year		= {1993},
  volume	= {140},
  number	= {3},
  pages		= {176--180},
  month		= {May},
  abstract	= {Variations on the standard Kohonen feature map can enable
		  an ordering of the map state space by using only a limited
		  subset of the complete input vector. Also it is possible to
		  employ merely a local adaptation procedure to order the
		  map, rather than having to rely on global variables and
		  objectives. Such variations have been included as part of a
		  hybrid learning system (HLS) which has arisen out of a
		  genetic-based classifier system. In this paper a
		  description of the modified feature map is given, which
		  constitutes the HLSs long term memory, and results on the
		  control of a simple maze running task are presented,
		  thereby demonstrating the value of goal related feedback
		  within the overall network.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball93b,
  author	= {Ball, N. R. },
  title		= {Towards the development of cognitive maps in classifier
		  systems},
  booktitle	= {Artificial Neural Nets and Genetic Algorithms. Proceedings
		  of the International Conference},
  year		= {1993},
  editor	= {Albrecht, R. F. and Reeves, C. R. and Steele, N. C. },
  pages		= {712--18},
  organization	= {Dept. of Eng. , Cambridge Univ. , UK},
  publisher	= {Springer-Verlag},
  address	= {Berlin, Germany},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball94a,
  author	= {N. R. Ball},
  title		= {Reinforcement Learning in {K}ohonen Feature Maps},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {I},
  pages		= {663--666},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, control, modification},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball94b,
  author	= {N. R. Ball},
  title		= {Application of a Neural Network based classifier system to
		  {ABV} obstacle avoidance},
  booktitle	= {Proc. IMACS Int. Symp. on Signal Processing, Robotics and
		  Neural Networks},
  year		= {1994},
  pages		= {294--297},
  publisher	= {IMACS},
  address	= {Lille, France},
  dbinsdate	= {oldtimer}
}

@InCollection{	  ball94c,
  author	= {N. Ball},
  title		= {Organizing an animat`s behavioural repertoires using
		  {K}ohonen feature maps},
  booktitle	= {From Animals to Animats 3. Proceedings of the Third
		  International Conference on Simulation of Adaptive
		  Behavior},
  publisher	= {MIT Press},
  year		= {1994},
  editor	= {D. Cliff and P. Husbands and J. -A. Meyer and S. W.
		  Wilson},
  address	= {Cambridge, MA, USA},
  pages		= {128--37},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  ball96a,
  author	= {N. R. Ball},
  title		= {Representation of obstacles in a Neural Network based
		  Classifier System},
  booktitle	= {Proc. ESANN'96, European Symp. on Artificial Neural
		  Networks},
  year		= {1996},
  publisher	= {D facto conference services},
  address	= {Bruges, Belgium},
  editor	= {Michel Verleysen},
  pages		= {155--160},
  dbinsdate	= {oldtimer}
}

@Article{	  ball96b,
  author	= {N. R. Ball},
  title		= {Application of a neural network based classifier system to
		  {AGV} obstacle avoidance},
  journal	= {Mathematics and Computers in Simulation},
  year		= {1996},
  volume	= {41},
  number	= {3--4},
  pages		= {285--96},
  note		= {(IMACS Symposium on Signal Processing Robotics and Neural
		  Networks Conference Date: April 1994 Conf. Loc: Lille,
		  France)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  balmat91a,
  author	= {Balmat, J. and Abellard, P. and Maifret, R. },
  title		= {Modeling {K}ohonen type neural networks using a data flow
		  Petri net},
  booktitle	= {Proceedings of the Fourth ISMM/IASTED International
		  Conference Parallel and Distributed Computing and
		  Systems---II},
  year		= {1991},
  editor	= {Ammar, R. A. },
  pages		= {32--4},
  organization	= {DCN Toulon, France},
  publisher	= {Acta Press},
  address	= {Anaheim, CA, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  balmelli96a,
  author	= {L. Balmelli},
  title		= {Adaptive sampling for very large particle systems using an
		  incremental \mbox{self-organizing} feature map: an
		  application in molecular dynamic},
  booktitle	= {Computer Animation and Simulation '96. Proceedings of the
		  Eurographics Workshop},
  publisher	= {Springer-Verlag},
  year		= {1996},
  editor	= {R. Boulic and G. Hegron},
  address	= {Wien, Austria},
  pages		= {15--29},
  dbinsdate	= {oldtimer}
}

@Article{	  ban00a,
  author	= {Ban, Sang-Woo and Cho, Jun-ki and Jung, Soon-ki and Lee,
		  Minho},
  title		= {Active vision system based on human eye saccadic
		  movement},
  journal	= {IEICE Transactions on Fundamentals of Electronics,
		  Communications and Computer Sciences},
  year		= {2000},
  volume	= {E83-A},
  number	= {6},
  month		= {Jun},
  pages		= {1066--1073},
  organization	= {Kyungpook Natl Univ},
  publisher	= {IEICE of Japan},
  address	= {Tokyo},
  abstract	= {We propose a new active vision system that mimics a
		  saccadic movement of human eye. It is implemented based on
		  a new computational model using neural networks. In this
		  model, the visual pathway was divided in order to
		  categorize a saccadic eye movement into three parts, each
		  of which was then individually modeled using different
		  neural networks to reflect a principal functionality of
		  brain structures related with the saccadic eye movement in
		  our brain. Initially, the visual cortex for saccadic eye
		  movements was modeled using a self-organizing feature map,
		  then a modified learning vector quantization network was
		  applied to imitate the activity of the superior colliculus
		  relative to a visual stimulus. In addition, a multilayer
		  recurrent neural network, which is learned by an
		  evolutionary computation algorithm, was used to model the
		  visual pathway from the superior colliculus to the
		  oculomotor neurons. Results from a computer simulation show
		  that the proposed computational model is effective in
		  mimicking the human eye movements during a saccade. Based
		  on the proposed model, an active vision system using a CCD
		  type camera and motor system was developed and demonstrated
		  with experimental results.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bandeira00a,
  author	= {Bandeira, N. and Lobo, V. and Moura Pires, F.},
  title		= {Analysis of {EEG} of shooters},
  booktitle	= {Proceedings of the International Conference on Mathematics
		  and Engineering Techniques in Medicine and Biological
		  Sciences. METMBS'00. CSREA Press---Univ. Georgia, Athens,
		  GA, USA},
  year		= {2000},
  volume	= {1},
  pages		= {35--9},
  abstract	= {Electroencephalogram (EEG) data was collected from
		  competition-level shooters during practice sessions and
		  searched for relations between this EEG data and the
		  shooter's score on target. For this purpose self-organizing
		  maps (SOM), cross-correlation analysis and scatter matrix
		  based techniques were applied to identify which variables
		  were relevant for the correct classification of EEG data
		  vectors into the corresponding target score classes.},
  dbinsdate	= {2002/1}
}

@InCollection{	  bandeira98a,
  author	= {N. Bandeira and V. J. Lobo and F. Moura-Pires},
  title		= {Training a \mbox{self-organizing} map distributed on a
		  {PVM} network},
  booktitle	= {1998 IEEE International Joint Conference on Neural
		  Networks Proceedings. IEEE World Congress on Computational
		  Intelligence},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {1},
  address	= {New York, NY, USA},
  pages		= {457--61},
  dbinsdate	= {oldtimer}
}

@Article{	  banzhaf90a,
  author	= {W. Banzhaf and H. Haken},
  title		= {Learning in a competitive network},
  journal	= {Neural Networks},
  year		= {1990},
  volume	= {3},
  number	= {4},
  pages		= {423--435},
  x		= {. . . The rule which was originally introduced by Kohonen
		  is appropriately modified and applied to the competitive
		  network under study. . . . },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baraghimian89a,
  author	= {K. A. Baraghimian},
  title		= {Connected component labeling using \mbox{self-organizing}
		  feature maps},
  booktitle	= {Proc. 13th Annual Int. Computer Software and Applications
		  Conference },
  year		= {1989},
  pages		= {680--684},
  publisher	= {IEEE Computer Soc. Press},
  address	= {Los Alamitos, CA},
  dbinsdate	= {oldtimer}
}

@Article{	  baraldi02a,
  author	= {Baraldi, A. and Alpaydin, E.},
  title		= {Constructive feedforward {ART} clustering networks---Part
		  {II}},
  journal	= {IEEE TRANSACTIONS ON NEURAL NETWORKS},
  year		= {2002},
  volume	= {13},
  number	= {3},
  month		= {MAY},
  pages		= {662--677},
  abstract	= {Part I of this paper defines the class of constructive
		  unsupervised on-line learning simplified adaptive resonance
		  theory (SART) clustering networks. Proposed instances of
		  class SART are the symmetric Fuzzy ART (S-Fuzzy ART) and
		  the Gaussian ART (GART) network. In Part 11 of our work, a
		  third network belonging to class SART, termed fully
		  self-organizing SART (FOSART), is presented and discussed.
		  FOSART is a constructive, soft-to-hard competitive,
		  topology-preserving, minimum- distance-to-means clustering
		  algorithm capable of: 1) generating processing units and
		  lateral connections on an example-driven basis and 2)
		  removing processing units and lateral connections on a
		  minibatch basis. FOSART is compared with Fuzzy ART, S-Fuzzy
		  ART, GART and other well-known clustering techniques (e.g.,
		  neural gas and self-organizing map) in several unsupervised
		  learning tasks, such as vector quantization, perceptual
		  grouping and 3-D surface reconstruction. These experiments
		  prove that when compared with other unsupervised learning
		  networks, FOSART provides an interesting balance between
		  easy user interaction, performance accuracy, efficiency,
		  robustness, and flexibility.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  baraldi95a,
  author	= {A. Baraldi and F. Parmiggiani},
  title		= {A Self-Organizing Neural Network Merging {K}ohonen's and
		  {ART} Models},
  volume	= {V},
  pages		= {2444--2449},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  baraldi95b,
  author	= {Baraldi, A. and Parmiggiani, F. },
  title		= {A neural network for unsupervised categorization of
		  multivalued input patterns: an application to satellite
		  image clustering},
  journal	= {IEEE Transactions on Geoscience and Remote Sensing},
  year		= {1995},
  volume	= {33},
  number	= {2},
  pages		= {305--16},
  month		= {March},
  dbinsdate	= {oldtimer}
}

@Article{	  baraldi96a,
  author	= {A. Baraldi and F. Parmiggiani},
  title		= {Fuzzy clustering: critical analysis of the contextual
		  mechanisms employed by three neural network models},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1996},
  volume	= {2761},
  pages		= {261--70},
  note		= {(Applications of Fuzzy Logic Technology III Conf. Date:
		  10--12 April 1996 Conf. Loc: Orlando, FL, USA Conf.
		  Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  baraldi97a,
  author	= {A. Baraldi and F. Parmiggiani},
  title		= {Neural network fuzzification: a critical review of the
		  fuzzy learning vector quantization model},
  booktitle	= {Neural Nets WIRN VIETRI-96. Proceedings of the 8th Italian
		  Workshop on Neural Nets},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {M. Marinaro and R. Tagliaferri},
  address	= {London, UK},
  pages		= {93--9},
  dbinsdate	= {oldtimer}
}

@Article{	  baraldi97b,
  author	= {A. Baraldi and F. Parmiggiani},
  title		= {Novel neural network model combining radial basis
		  function, competitive Hebbian learning rule, and fuzzy
		  simplified adaptive resonance theory},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1997},
  volume	= {3165},
  pages		= {98--112},
  note		= {(Applications of Soft Computing Conf. Date: 28--29 July
		  1997 Conf. Loc: San Diego, CA, USA Conf. Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  baraldi97c,
  author	= {A. Baraldi and F. Parmiggiani},
  title		= {Fuzzy Combination of {K}ohonen's and {ART} Neural Network
		  Models to Detect Statistical Regularities in a Random
		  Sequence of Multi-Valued Input Patterns},
  booktitle	= {Proceedings of ICNN'97, International Conference on Neural
		  Networks},
  publisher	= {IEEE Service Center},
  year		= 1997,
  address	= {Piscataway, NJ},
  volume	= {I},
  pages		= {281--286},
  dbinsdate	= {oldtimer}
}

@InCollection{	  baraldi98a,
  author	= {A. Baraldi and P. Blonda and A. Petrosino},
  title		= {Fuzzy neural networks for pattern recognition},
  booktitle	= {Neural Nets WIRN-VIETRI-97. Proceedings of the 9th Italian
		  Workshop on Neural Nets},
  publisher	= {Springer-Verlag London},
  year		= {1998},
  editor	= {M. Marinaro and R. Tagliaferri},
  address	= {London, UK},
  pages		= {35--83},
  dbinsdate	= {oldtimer}
}

@Article{	  baraldi98b,
  author	= {Baraldi, A. and Blonda, P. and Parmiggiani, F. and
		  Pasquariello, G. and Satalino, G.},
  title		= {Model transitions in descending {F LVQ} },
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1998},
  number	= {5},
  volume	= {9},
  pages		= {724--738},
  abstract	= {Fuzzy learning vector quantization (FLVQ), also known as
		  the fuzzy Kohonen clustering network, was developed to
		  improve performance and usability of on-line
		  hard-competitive Kohonen's vector quantization and
		  soft-competitive self organizing map (SOM) algorithms. The
		  FLVQ effectiveness seems to depend on the range of change
		  of the weighting exponent m(t). In the first part of this
		  work, extreme m(t) values (1 and infinity , respectively)
		  are employed to investigate FLVQ asymptotic behaviors. This
		  analysis shows that when m(t) tends to either one of its
		  extremes, FLVQ is affected by trivial vector quantization,
		  which causes centroids collapse in the grand mean of the
		  input data set. No analytical criterion has been found to
		  improve the heuristic choice of the range of m(t) change.
		  In the second part of this paper, two FLVQ and {SOM}
		  classification experiments of remote sensed data are
		  presented. In these experiments the two nets are connected
		  in cascade to a supervised second stage, based on the delta
		  rule. Experimental results confirm that FLVQ performance
		  can be greatly affected by the user's definition of the
		  range of change of the weighting exponent. Moreover, FLVQ
		  shows instability when its traditional termination
		  criterion is applied. Empirical recommendations are
		  proposed for the enhancement of FLVQ robustness. Both the
		  analytical and the experimental data reported seem to
		  indicate that the choice of the range of m(t) change is
		  still open to discussion and that alternative clustering
		  neural-network approaches should be developed to pursue
		  during training: 1) monotone reduction of the neurons'
		  learning rate and 2) monotone reduction of the overlap
		  among neuron receptive fields.},
  dbinsdate	= {oldtimer}
}

@Article{	  baraldi99a,
  author	= {Baraldi, A. and Blonda, P.},
  title		= {A survey of fuzzy clustering algorithms for pattern
		  recognition. {II}},
  journal	= {IEEE Transactions on Systems, Man and Cybernetics, Part B
		  (Cybernetics)},
  year		= {1999},
  volume	= {29},
  pages		= {786--801},
  abstract	= {For pt.I see ibid., p.775--85. In part I an equivalence
		  between the concepts of fuzzy clustering and soft
		  competitive learning in clustering algorithms is proposed
		  on the basis of the existing literature. Moreover, a set of
		  functional attributes is selected for use as dictionary
		  entries in the comparison of clustering algorithms. In this
		  paper, five clustering algorithms taken from the literature
		  are reviewed, assessed and compared on the basis of the
		  selected properties of interest. These clustering models
		  are (1) self-organizing map (SOM); (2) fuzzy learning
		  vector quantization (FLVQ); (3) fuzzy adaptive resonance
		  theory (fuzzy ART); (4) growing neural gas (GNG); (5) fully
		  self-organizing simplified adaptive resonance theory
		  (FOSART). Although our theoretical comparison is fairly
		  simple, it yields observations that may appear parodoxical.
		  First, only FLVQ, fuzzy ART, and FOSART exploit concepts
		  derived from fuzzy set theory (e.g., relative and/or
		  absolute fuzzy membership functions). Secondly, only {SOM},
		  FLVQ, GNG, and FOSART employ soft competitive learning
		  mechanisms, which are affected by asymptotic misbehaviors
		  in the case of FLVQ, i.e., only {SOM}, GNG, and FOSART are
		  considered effective fuzzy clustering algorithms.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baras90a,
  author	= {J. S. Baras and A. LaVigna},
  title		= {Convergence of {K}ohonen's learning vector quantization},
  booktitle	= {Proc. IJCNN-90, International Joint Conference on Neural
		  Networks, San Diego},
  year		= {1990},
  volume	= {III},
  pages		= {17--20},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baras90b,
  author	= {J. S. Baras and A. LaVigna},
  title		= {Convergence of the vectors in {K}ohonen's learning vector
		  quantization},
  booktitle	= {Proc. INNC'90, Int. Neural Network Conference },
  year		= {1990},
  volume	= {II},
  pages		= {1028--1031},
  organization	= {Thomsom; SUN; British Computer Society ; et al},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baras90c,
  author	= {J. S. Baras and A. La Vigna},
  title		= {Convergence of a neural network classifier},
  booktitle	= {Proc. 29th IEEE Conference on Decision and Control},
  year		= {1990},
  volume	= {III},
  pages		= {1735--1740},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baras99a,
  author	= {Baras, J. S. and Dey, S.},
  title		= {Adaptive classification based on compressed data using
		  learning vector quantization},
  booktitle	= {Proceedings of the 38th IEEE Conference on Decision and
		  Control},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {4},
  pages		= {3677--83},
  abstract	= {Classification problems using compressed data are becoming
		  increasingly important in many applications with large
		  amounts of sensory data and large sets of classes. These
		  applications range from aided target recognition (ATR), to
		  medical diagnosis, to speech recognition, to fault
		  detection and identification in manufacturing systems. In
		  this paper, we develop and analyze a learning vector
		  quantization (LVQ) based algorithm for the combined
		  compression and classification problem. We show convergence
		  of the algorithm using techniques from stochastic
		  approximation, namely, the ODE method.},
  dbinsdate	= {oldtimer}
}

@Article{	  baras99b,
  author	= {Baras, J. S. and Dey, S.},
  title		= {Combined compression and classification with learning
		  vector quantization},
  journal	= {IEEE Transactions on Information Theory},
  year		= {1999},
  volume	= {45},
  pages		= {1911--20},
  abstract	= {Combined compression and classification problems are
		  becoming increasingly important in many applications with
		  large amounts of sensory data and large sets of classes.
		  These applications range from automatic target recognition
		  (ATR) to medical diagnosis, speech recognition, and fault
		  detection and identification in manufacturing systems. In
		  this paper, we develop and analyze a learning vector
		  quantization (LVQ) based algorithm for combined compression
		  and classification. We show convergence of the algorithm
		  using the ODE method from stochastic approximation. We
		  illustrate the performance of the algorithm with some
		  examples.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  barbalho01a,
  author	= {Barbalho, J. M. and Neto, A. D. D. and Costa, J. A. E. and
		  Netto, M. L. A.},
  title		= {Hierarchical {SOM} applied to image compression},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2001},
  editor	= {},
  volume	= {1},
  pages		= {442--447},
  organization	= {Univ. Federal do Rio Grande do Norte, Dept. of Electrical
		  Engineering, Lab. of Computer Eng. and Automation},
  publisher	= {},
  address	= {},
  abstract	= {The increase of the need for image storage and
		  transmission in computer systems has increased the
		  importance of signal and image compression algorithms. The
		  approach involving vector quantization (VQ) relies on
		  designing of a finite set of codes which will substitute
		  the original signal during transmission with a minimal of
		  distortion, taking advantage of the spatial redundancy of
		  image to compress them. Algorithms such as LGB and SOM work
		  in an unsupervised way toward finding a good codebook for a
		  given training data. However, the number of code vectors
		  (N) needed for VQ increases with the vector dimension, and
		  full-search algorithms such as LGB and SOM can lead to
		  large training and coding times. An alternative for
		  reducing the computational complexity is the use of a
		  tree-structured vector quantization algorithm. This paper
		  presents an application of a hierarchical SOM for image
		  compression in which reduces the search complexity from
		  O(N) to O(log N), enabling a faster training and image
		  coding. Results are given for conventional SOM, LBG and
		  HSOM showing the advantage of the proposed method.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  barber93a,
  author	= {Steven M. Barber and Jose G. Delgado-Frias and Stamatis
		  Vassiliadis and Gerald G. Pechanek},
  title		= {{SPIN-L}: Sequential Pipelined Neuroemulator with Learning
		  Capabilities},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {II},
  pages		= {1927--1930},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  barea98a,
  author	= {Barea, R. and Boquete, L. and Garcia, R. and Mazo, M. and
		  Lopez, E. and Escudero M. and Rodriguez F. J.},
  title		= {{ECG} identification using self-organising maps},
  booktitle	= {Proceedings of NC 1998. International ICSC/IFAC Symposium
		  on Neural Computation. ICSC Academic Press, Zurich,
		  Switzerland},
  year		= {1998},
  volume	= {},
  pages		= {378--83},
  abstract	= {The use of automatic systems for the analysis of ECG
		  signals has been growing since the early sixties due to the
		  possibilities opened up by the new technologies. This paper
		  presents the results of the use of self-organising Kohonen
		  maps for the automatic identification of the various
		  cardiac arrhythmias. The algorithm proposed has two
		  options: automatic identification of the R wave in the QRS
		  complex and classification the static projection of the
		  cardiac cycle. Tests carried out with synthetic,
		  noise-contaminated signals suggest that the system could be
		  useful, though further corroborative tests need to be
		  carried out with real signals.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  barge95a,
  author	= {Barge, M. and Chevallier, R. and Curatu, E. and Maruani,
		  A. },
  title		= {Optical digit recognition based on {K}ohonen maps},
  booktitle	= {Optical Computing. Proceedings of the International
		  Conference},
  year		= {1995},
  editor	= {Wherrett, B. S. and Chavel, P. },
  pages		= {451--4},
  organization	= {Dept. Images, ENST, Paris, France},
  publisher	= {IOP Publishing},
  address	= {Bristol, UK},
  dbinsdate	= {oldtimer}
}

@Article{	  barge96a,
  author	= {M. Barge and K. Heggarty and Y. Idan and R. Chevallier},
  title		= {64-channel correlator implementing a {K}ohonen-like neural
		  network for handwritten-digit recognition},
  journal	= {Applied Optics},
  year		= {1996},
  volume	= {35},
  number	= {23},
  pages		= {4655--65},
  dbinsdate	= {oldtimer}
}

@Article{	  barhak01a,
  author	= {Barhak, J. and Fischer, A.},
  title		= {Parameterization and reconstruction from 3D scattered
		  points based on neural network and {PDE} techniques},
  journal	= {IEEE Transactions on Visualization and Computer Graphics},
  year		= {2001},
  volume	= {7},
  number	= {1},
  month		= {January/March 2001},
  pages		= {1--16},
  organization	= {CMSR Lab. for Comp. Graphics and CAD, Department of
		  Mechanical Engineering, Technion},
  publisher	= {},
  address	= {},
  abstract	= {Reverse engineering ordinarily uses laser scanners since
		  they can sample 3D data quickly and accurately relative to
		  other systems. These laser scanner systems, however, yield
		  an enormous amount of irregular and scattered digitized
		  point data that requires intensive reconstruction
		  processing. Reconstruction of freeform objects consists of
		  two main stages: 1) parameterization and 2) surface
		  fitting. Selection of an appropriate parameterization is
		  essential for topology reconstruction as well as surface
		  fitness. Current parameterization methods have topological
		  problems that lead to undesired surface fitting results,
		  such as noisy self-intersecting surfaces. Such problems are
		  particularly common with concave shapes whose parametric
		  grid is self-intersecting, resulting in a fitted surface
		  that considerably twists and changes its original shape. In
		  such cases, other parameterization approaches should be
		  used in order to guarantee non-self-intersecting behavior.
		  The parameterization method described in this paper is
		  based on two stages: 1) 2D initial parameterization and 2)
		  3D adaptive parameterization. Two methods were developed
		  for the first stage: Partial Differential Equation (PDE)
		  parameterization and neural network Self Organizing Maps
		  (SOM) parameterization. PDE parameterization yields a
		  parametric grid without self-intersections. Neural network
		  SOM parameterization creates a grid where all the sampled
		  points, not only the boundary points, affect the grid,
		  leading to a uniform and smooth surface. In the second
		  stage, a 3D base surface was created and then adaptively
		  modified. To this end, the Gradient Descent Algorithm (GDA)
		  and Random Surface Error Correction (RSEC), both of which
		  are iterative surface fitting methods, were developed and
		  implemented. The feasibility of the developed
		  parameterization methods and fitting algorithms is
		  demonstrated on several examples using sculptured free
		  objects.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  barhak01b,
  author	= {Barhak, J. and Fischer, A.},
  title		= {Adaptive reconstruction of freeform objects with 3D {SOM}
		  neural network grids},
  booktitle	= {Proceedings Ninth Pacific Conference on Computer Graphics
		  and Applications. Pacific Graphics 2001. IEEE Comput. Soc,
		  Los Alamitos, CA, USA},
  year		= {2001},
  volume	= {},
  pages		= {97--105},
  abstract	= {Reverse engineering is an important process in CAD systems
		  today. Yet several open problems lead to a bottleneck in
		  the reverse engineering process. First, because the
		  topology of the object to be reconstructed is unknown,
		  point connectivity relations are undefined. Second, the
		  fitted surface must satisfy global and local shape
		  preservation criteria that are undefined explicitly. In
		  reverse engineering, object reconstruction is based both on
		  parameterization and on fitting. Nevertheless, the above
		  problems are influenced mainly by parameterization. In
		  order to overcome the above problems, the paper proposes a
		  neural network, Self Organizing Map (SOM) method, for
		  creating a 3D parametric grid. The main advantage of the
		  proposed SOM method is that it detects both the orientation
		  of the grid and the position of the sub-boundaries. The
		  neural network grid converges to the sampled shape through
		  an adaptive learning process. The SOM method is applied
		  directly on 3D sampled data and avoids the projection
		  anomalies common to other methods. The paper also presents
		  boundary correction and growing grid extensions to the SOM
		  method. In the surface fitting stage, an RSEC (Random
		  Surface Error Correction) fitting method based on the SOM
		  method was developed and implemented.},
  dbinsdate	= {2002/1}
}

@MastersThesis{	  barmore88a,
  author	= {G. D. Barmore},
  title		= {Speech Recognition Using Neural Nets and Dynamic Time
		  Warping},
  school	= {Air Force Inst. of Tech. },
  year		= {1988},
  address	= {Wright-Patterson AFB, OH},
  month		= {December},
  dbinsdate	= {oldtimer}
}

@TechReport{	  barna87a,
  author	= {G. Barna},
  title		= {Modification of {K}ohonen's Self-Organizing Algorithm:
		  Numerical Studies},
  institution	= {Helsinki Univ. of Technology, Lab. of Computer and
		  Information Science},
  year		= {1987},
  type		= {Report {A4}},
  address	= {Espoo, Finland},
  month		= {October},
  dbinsdate	= {oldtimer}
}

@Article{	  barna88a,
  author	= {Gy{\"{o}}rgy Barna and Ronald Chrisley and Teuvo Kohonen},
  title		= {Statistical Practical Pattern Recognition with Neural
		  Networks},
  journal	= {Neural Networks},
  year		= {1988},
  volume	= {1},
  number	= {Supplement 1},
  pages		= {7},
  annotate	= {},
  abstract	= {Successful recognition of natural signals, e.g., speech,
		  requires substantial statistical pattern recognition
		  capabilities. It is thus desirable to compare the
		  performance of the neural approaches with each other and to
		  the performance of more conventional approaches on tasks
		  that are explicitly statistical. Although (Huang and
		  Lippmann, l987) compares a Backpropagation network (BP) to
		  conventional Bayesian classifiers on some tasks that have
		  some statistical aspects, this work is novel in several
		  respect: 1) tasks of varying dimensionality are considered;
		  2) none of the tasks include a uniform distribution
		  component that, although popular in non-statistical pattern
		  recognition tasks, is rarely encountered in natural
		  applications; and 3) several networks are compared: a BP,
		  two Boltzmann machines (BM1 and BM2), and the Learning
		  Vector Quantization method (LVQ), developed previously by
		  one of the authors (Kohonen, l988).},
  dbinsdate	= {oldtimer}
}

@Article{	  barna89a,
  author	= {G. Barna and K. Kaski},
  title		= {Variations on the {B}oltzmann machine},
  journal	= {J. Physics A [Mathematical and General]},
  year		= {1989},
  volume	= {22},
  number	= {23},
  pages		= {5174--5151},
  dbinsdate	= {oldtimer}
}

@Article{	  barna90a,
  author	= {G. Barna and K. Kaski},
  title		= {Stochastic vs. deterministic neural networks for pattern
		  recognition},
  journal	= {Physica Scripta},
  year		= {1990},
  volume	= {T33},
  pages		= {110--115},
  note		= {Conference paper in journal},
  dbinsdate	= {oldtimer}
}

@Article{	  barnickel97a,
  author	= {Barnickel,G. and Anzali,S},
  title		= {Evaluation of High Throughput Screening Hits by Means of
		  {K}ohonen Neural Networks},
  journal	= {Abstr. Pap. Amer. Chem. Soc. },
  year		= {1997},
  pages		= {29-?},
  volume	= {214},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  barone95a,
  author	= {Dante Augusto Couto Barone and Ant{\^{o}}nio Rog{\`{e}}rio
		  Machado Ramos},
  title		= {Application of a Hybrid System in Engineering Pattern
		  Recognition Problems},
  booktitle	= {Proc. EANN'95, Engineering Applications of Artificial
		  Neural Networks},
  year		= {1995},
  pages		= {95--98},
  organization	= {Finnish Artificial Intelligence Society},
  dbinsdate	= {oldtimer}
}

@Article{	  barragan00a,
  author	= {Barragan, M. J. R. and Bote, V. P. G. and Guerrero, A. P.
		  and Alonso, F. Z.},
  title		= {Scientific journals: determination of the requirements of
		  use},
  journal	= {Revista-Espanola-de-Documentacion-Cientifica},
  year		= {2000},
  volume	= {23},
  pages		= {417--36},
  abstract	= {The information requirements of the scientific community
		  of the areas of science and technology in the University of
		  Extremadura were analysed quantitatively and qualitatively
		  through journal citations. For this purpose, a
		  retrospective search was made in the Science Citation Index
		  database, considering multiple variables such as use by
		  department, impact factor, journal citations, availability,
		  localization, accessibility, and coverage. A set of (simple
		  and complex) indicators was applied to these variables,
		  using multiple association models (in particular, Kohonen's
		  algorithm) to determine the degree of interdepartmental
		  overlap in the real use of information.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  barreto01a,
  author	= {Barreto, G. A. and Araujo, A. F. R.},
  title		= {A self-organizing {NARX} network and its application to
		  prediction of chaotic time series},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2001},
  editor	= {},
  volume	= {3},
  pages		= {2144--2149},
  organization	= {Depto. de Engenharia Electrica USP},
  publisher	= {},
  address	= {},
  abstract	= {This paper introduces the concept of dynamic embedding
		  manifold (DEM), which allows the Kohonen self-organizing
		  map (SOM) to learn dynamic, nonlinear input-output
		  mappings. The combination of the DEM concept with the SOM
		  results in a new modelling technique that we called
		  Vector-Quantized Temporal Association Memory (VQTAM). We
		  use VQTAM to propose an unsupervised neural algorithm
		  called Self-Organizing NARX (SONARX) network. The SONARX
		  network is evaluated on the problem of modeling and
		  prediction of three chaotic time series and compared with
		  MLP, RBF and autoregressive (AR) models. Its is shown that
		  SONARX exhibits similar performance when compared to MLP
		  and RBF, while producing much better results than the AR
		  model. The influence of the number of neurons, the memory
		  order, the number of training epochs and the size of the
		  training set in the final prediction error is also
		  evaluated.},
  dbinsdate	= {2002/1}
}

@Article{	  barreto01b,
  author	= {G. de A. Barreto and Aluizio F. R. Ara{\'u}jo},
  title		= {Time in Self-Organizing Maps: An Overview of Models},
  journal	= {International Journal of Computer Research, Special Issue:
		  Past, Present and Future of Neural Networks},
  year		= {2001},
  key		= {},
  volume	= {10},
  number	= {2},
  pages		= {139--79},
  month		= {},
  note		= {Guest Editors: P.G. Anderson and G. Antoniou and V.
		  Mladenov and E. Oja and M. Paprzycki and N. C. Steele},
  annote	= {},
  dbinsdate	= {2002/1}
}

@InProceedings{	  barritt92a,
  author	= {Barritt, B. A. and Rau, N. J. },
  title		= {Enhancing electronic combat system digital signal
		  processing using neural networks},
  booktitle	= {Proceedings of the IEEE 1992 National Aerospace and
		  Electronics Conference, NAECON 1992},
  year		= {1992},
  volume	= {3},
  pages		= {887--93},
  organization	= {Logicon Inc. , Dayton, OH, USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  barro93a,
  author	= {S. Barro and M. G. Penedo and D. Cabello and J. M. Pardo},
  title		= {Artificial neural network based processing in a system for
		  lung nodule detection},
  booktitle	= {Conference Proceedings DICTA-93 Digital Image Computing:
		  Techniques and Applications},
  publisher	= {Australian Pattern Recognition Soc},
  year		= {1993},
  volume	= {1},
  editor	= {K. K. Fung and A. Ginige},
  address	= {Broadway, NSW, Australia},
  pages		= {79--86},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  barry93a,
  author	= {William Barry and Paul Dalsgaard},
  title		= {Speech Database Annotation. The Importance of a
		  Multi-Lingual Approach},
  booktitle	= {Proc. EUROSPEECH'93, 3rd European Conference on Speech,
		  Communication and Technology},
  year		= {1993},
  volume	= {I},
  pages		= {13--20},
  address	= {Berlin},
  dbinsdate	= {oldtimer}
}

@InCollection{	  barschdorff94a,
  author	= {D. Barschdorff and U. Femmer},
  title		= {Artificial neural networks for wear estimation},
  booktitle	= {Intelligent Manufacturing Systems 1994 (IMS`94). A
		  Postprint Volume from the IFAC Workshop},
  publisher	= {Pergamon},
  year		= {1994},
  editor	= {P. Kopacek},
  address	= {Oxford, UK},
  pages		= {151--5},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  barson95a,
  author	= {P. Barson and N. Davey and S. Field and R. Frank and D. S.
		  W. Tansley},
  title		= {Dynamic Competitive Learning Applied to the Clone
		  Detection Problem},
  booktitle	= {Proc. Int. Workshop on Applications of Neural Networks to
		  Telecommunications 2},
  year		= {1995},
  editor	= {Joshua Alspector and Rodney Goodman and Timothy X. Brown},
  pages		= {234--241},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bartal94a,
  author	= {Yair Bartal and Jie Lin and Robert E. Uhrig},
  title		= {Nuclear Power Plant Transient Diagnostics Using {LVQ} or
		  {SOM} networks don't know that they don't know},
  pages		= {3744--3749},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, diagnostics, monitoring},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baruah91a,
  author	= {Arati B. Baruah and Les E. Atlas and Alistair D. C.
		  Holden},
  title		= {{K}ohonen's feature maps applied to ordered clustering
		  applications},
  booktitle	= {Proc. IJCNN'91, International Joint Conference on Neural
		  Networks},
  year		= {1991},
  volume	= {I},
  pages		= {596--601},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bastos99a,
  author	= {Bastos, R. C. and Bastos, L. C.},
  title		= {Success factors determination for entrepreneurs in Santa
		  Catarina by using neural networks},
  booktitle	= {IJCNN'99. International Joint Conference on Neural
		  Networks. Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {5},
  pages		= {3556--9},
  abstract	= {The entrepreneurs has a fundamental role in the success of
		  an undertaking. Besides the features and attributes of
		  their personality, the attitude they take has an important
		  influence in the undertaking success. This paper presents
		  the neural networks utilization as a tool directed to
		  investigate the existence of patterns for successful
		  entrepreneurs defined according to a set of features. The
		  use of neural networks makes possible the classification
		  and recognition of patterns associated to the entrepreneurs
		  success or failure. From real data about entrepreneurs in
		  Santa Catarina State, a set of LVQ neural network
		  architectures is tested in order to establish a
		  classification model of successful entrepreneurs. The
		  creation of profiles which allow relevant feature
		  evaluation in the identification of a successful
		  entrepreneur is also discussed.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bastos99b,
  author	= {Bastos, L. C. and Bastos, R. C. and Nishida, W.},
  title		= {Radial basis function for classification of remote sensing
		  images},
  booktitle	= {IJCNN'99. International Joint Conference on Neural
		  Networks. Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {3},
  pages		= {1959--62},
  abstract	= {This work presents a hybrid classifier for multispectral
		  images using radial basis function networks (RBF). A
		  Kohonen self-organization-map is used in substitution of
		  the k-means algorithm in unsupervised stage of training.
		  The algorithm of the pseudo-inverse is used for the
		  determination of the weights of the supervised stage. The
		  architecture proposed reduces the time required for
		  processing. Also, it presents satisfactory results with
		  small training samples. A practical application is
		  accomplished and the results obtained between the
		  classifier of maximum likelihood and the proposed hybrid
		  classifier are compared.},
  dbinsdate	= {oldtimer}
}

@Article{	  bathen00a,
  author	= {Bathen, T. F. and Engan, T. and Krane, J. and Axelson,
		  D.},
  title		= {Analysis and classification of proton {NMR} spectra of
		  lipoprotein fractions from healthy volunteers and patients
		  with cancer or {CHD}},
  journal	= {ANTICANCER RESEARCH},
  year		= {2000},
  volume	= {20},
  number	= {4},
  month		= {JUL-AUG},
  pages		= {2393--2408},
  abstract	= {Human blood plasma samples from 52 subjects were collected
		  and the very low density lipoprotein (VLDL) intermediate
		  density lipoprotein (IDL) low density lipoprotein (LDL) and
		  high density lipoprotein were isolated by serial ultra
		  centrifugation. 600 MHz H-1 NMR spectra of the lipoprotein
		  fractions were acquired. The methyl and methylene regions
		  in the spectra of VLDL, LDL and HDL were utilised in
		  further analyses via Kohonen neural networks (KNN) and
		  generative topographic mapping (GTM) two related examples
		  of (unsupervised learning) self-organising feature mapping
		  techniques. Systematic variations in lipoprotein profiles
		  can be substantially visualised through the use of KNN and
		  GTM. The relationship between the sample positions in the
		  Kohonen plot was visualised by surface plots of the
		  corresponding VLDL and HDL cholesterol and VLDL
		  triglyceride contents. The GTM maps of the VLDL and HDL
		  fractions were used to investigate the individual
		  properties of selected samples. A large number of the
		  cancer patients were found clustered in the VLDL GTM map,
		  and GTM map positions of samples in relation to CHD,
		  diabetes and renal failure could be found. Although the
		  study group here considered is heterogeneous in respect to
		  age, sex, type of disease and medications within each
		  defined class, classification of VLDL and HDL data with
		  probabilistic neural network (PNN) was quite successful
		  with respect to the groupings: cancer, CHD, volunteers and
		  other (comprising patients with other diseases). Statistics
		  based on 15 independent sets of PNN calculations gave true
		  positive fractions usually higher than 0.83 and false
		  positive fractions lower than 0.088. Attempts to use the
		  corresponding LDL data and four classes were uniformly poor
		  although some classifications (e.g, volunteer versus CHD)
		  were easily performed.},
  dbinsdate	= {2002/1}
}

@Article{	  bauer92a,
  author	= {Hans-Ulrich Bauer and Klaus R. Pawelzik},
  title		= {Quantifying the neighborhood preservation of
		  {S}elf-{O}rganizing {F}eature {M}aps},
  journal	= {IEEE Trans. on Neural Networks},
  year		= {1992},
  volume	= {3},
  number	= {4},
  pages		= {570--579},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bauer92b,
  author	= {Hans-Ulrich Bauer and Klaus Pawelzik and Theo Geisel},
  title		= {A Topographic Product for the Optimization of
		  Self-Organizing Feature Maps},
  booktitle	= {Advances in Neural Information Processing Systems 4},
  editor	= {John E. Moody and Stephen J. Hanson and Richard P.
		  Lippmann},
  publisher	= {Morgan Kaufmann},
  year		= {1992},
  pages		= {1141--1147},
  address	= {San Mateo, CA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bauer94a,
  author	= {H. -U. Bauer},
  title		= {Oriented Ocular Dominance Bands in the {S}elf-{O}rganizing
		  {F}eature {M}ap},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {I},
  pages		= {42--45},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {analysis, visual system},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bauer95a,
  author	= {H. -U. Bauer and Th. Villmann},
  title		= {A Growth Algorithm for Hypercubical Output Spaces in
		  Self-Organizing Feature Maps},
  booktitle	= {Proc. ICANN'95, International Conference on Artificial
		  Neural Networks},
  year		= {1995},
  editor	= {F. Fogelman-Souli{\'{e}} and P. Gallinari},
  volume	= {I},
  pages		= {69--74},
  publisher	= {EC2},
  address	= {Nanterre, France},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer95b,
  author	= {Bauer, H. -U. },
  title		= {Development of oriented ocular dominance bands as a
		  consequence of areal geometry},
  journal	= {Neural Computation},
  year		= {1995},
  volume	= {7},
  number	= {1},
  pages		= {36--50},
  month		= {Jan},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer96a,
  author	= {H. -U. Bauer and R. Der and M. Herrmann},
  title		= {Controlling the magnification factor of
		  \mbox{self-organizing} feature maps},
  journal	= {Neural Computation},
  year		= {1996},
  volume	= {8},
  number	= {4},
  pages		= {757--71},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer96b,
  author	= {H. -U. Bauer and M. Riesenhuber and T. Geisel},
  title		= {Phase diagrams of \mbox{self-organizing} maps},
  journal	= {Physical Review E [Statistical Physics, Plasmas, Fluids,
		  and Related Interdisciplinary Topics]},
  year		= {1996},
  volume	= {54},
  number	= {3},
  pages		= {2807--10},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer97a,
  author	= {H. -U. Bauer and T. Villmann},
  title		= {Growing a hypercubical output space in a
		  \mbox{self-organizing} feature map},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1997},
  volume	= {8},
  number	= {2},
  pages		= {218--26},
  abstract	= {Neural maps project data from an input space onto a neuron
		  position in a (often lower dimensional) output space grid
		  in a neighborhood preserving way, with neighboring neurons
		  in the output space responding to neighboring data points
		  in the input space. A map-learning algorithm can achieve an
		  optimal neighborhood preservation only, if the output space
		  topology roughly matches the effective structure of the
		  data in the input space. We here present a growth
		  algorithm, called the GSOM or growing self-organizing map,
		  which enhances a widespread map self-organization process,
		  Kohonen's self-organizing feature map (SOFM), by an
		  adaptation of the output space grid during learning. The
		  GSOM restricts the output space structure to the shape of a
		  general hypercubical shape, with the overall dimensionality
		  of the grid and its extensions along the different
		  directions being subject of the adaptation. This constraint
		  meets the demands of many larger information processing
		  systems, of which the neural map can be a part. We apply
		  our GSOM-algorithm to three examples, two of which involve
		  real world data. Using recently developed methods for
		  measuring the degree of neighborhood preservation in neural
		  maps, we find the GSOM-algorithm to produce maps which
		  preserve neighborhoods in a nearly optimal fashion.},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer97b,
  author	= {H. -U. Bauer and D. Brockmann and T. Geisel},
  title		= {Analysis of ocular dominance pattern formation in a
		  high-dimensional \mbox{self-organizing}-map model},
  journal	= {Network: Computation in Neural Systems},
  year		= {1997},
  volume	= {8},
  number	= {1},
  pages		= {17--33},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bauer97c,
  author	= {H. -U. Bauer and M. Riesenhuber and D. Brockmann and T.
		  Geisel},
  title		= {Analysis of {SOM}-based models for the development of
		  visual maps},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {233--238},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer97d,
  author	= {H. U. Bauer and W. Sch{\"o}llhorn},
  title		= {Self-Organizing Maps for the Analysis of Complex Movement
		  Patterns},
  journal	= {Neural Processing Letters},
  year		= 1997,
  volume	= 5,
  pages		= {193--199},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer98a,
  author	= {Bauer, H. U.},
  title		= {Exploiting topography of neural maps: A case study on
		  investment strategies for emerging markets},
  journal	= {IEEE IAFE Conference on Computational Intelligence for
		  Financial Engineering (CIFEr)},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1998},
  number	= {},
  volume	= {},
  pages		= {216--219},
  abstract	= {Investments can be spread over many possible assets to
		  avoid risk (at the cost of obtaining only an average
		  performance) or it can be focused on clusters of only a few
		  promising assets (at the cost of increased risk). A
		  trade-off between these two objectives can be reached by
		  using the Self-Organizing Map (SOM), a neural network
		  paradigm which achieves a clustering of data points while
		  simultaneously preserving their inherent neighborhood
		  relations (topography). This amounts to a combination of
		  clustering with local smoothing. In a case study involving
		  investments in emerging stock markets I illustrate the
		  application of {SOM}s in investment decisions, with an
		  improvement of about 30% in returns over other, more simple
		  investment strategies.},
  dbinsdate	= {oldtimer}
}

@Article{	  bauer99a,
  author	= {Bauer, H. U and Herrmann, M. and Villmann, T.},
  title		= {Neural maps and topographic vector quantization},
  journal	= {Neural Networks},
  year		= {1999},
  number	= {4},
  volume	= {12},
  pages		= {659--676},
  abstract	= {Neural maps combine the representation of data by codebook
		  vectors, like a vector quantizer, with the property of
		  topography, like a continuous function. While the
		  quantization error is simple to compute and to compare
		  between different maps, topography of a map is difficult to
		  define and to quantify. Yet, topography of a neural map is
		  an advantageous property, e.g. in the presence of noise in
		  a transmission channel, in data visualization, and in
		  numerous other applications. In this article we review some
		  conceptual aspects of definitions of topography, and some
		  recently proposed measures to quantify topography. We apply
		  the measures first to neural maps trained on synthetic data
		  sets, and check the measures for properties like
		  reproducibility, scalability, systematic dependence of the
		  value of the measure on the topology of the map, etc. We
		  then test the measures on maps generated for four
		  real-world data sets, a chaotic time series, speech data,
		  and two sets of image data. The measures are found to do an
		  imperfect, but an adequate job in selecting a
		  topographically optimal output space dimension, while they
		  consistently single out particular maps as
		  non-topographic.},
  dbinsdate	= {oldtimer}
}

@Article{	  bauknecht96a,
  author	= {H. Bauknecht and A. Zell and H. Bayer and P. Levi and M.
		  Wagner and J. Sadowski and J. Gasteiger},
  title		= {Locating biologically active compounds in medium-sized
		  heterogeneous datasets by topical autocorrelation vectors:
		  Dopamine and benzodiazepine agonists},
  journal	= {Journal of Chemical Information and Computer Sciences},
  year		= 1996,
  volume	= 36,
  pages		= {1205--1213},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baumann91a,
  author	= {Thomas Baumann and Alain Germond and Daniel Tschudi},
  title		= {Impulse Test Fault Diagnosis on Power Transformers Using
		  {K}ohonen's Self-Organizing Neural Network},
  booktitle	= {Proc. Third Symp. on Expert Systems Application to Power
		  Systems},
  year		= {1991},
  address	= {Tokyo \& Kobe},
  dbinsdate	= {oldtimer}
}

@Article{	  baumann93a,
  author	= {Baumann, E. W. and Williams, D. L. },
  title		= {Stochastic associative memory},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1993},
  volume	= {1966},
  pages		= {132--9},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baumann93b,
  author	= {Baumann, T. and Germond, A. J. },
  title		= {Application of the {K}ohonen network to short-term load
		  forecasting},
  booktitle	= {ANNPS '93. Proceedings of the Second International Forum
		  on Applications of Neural Networks to Power Systems},
  year		= {1993},
  editor	= {Tamura, Y. and Suzuki, H. and Mori, H. },
  pages		= {407--12},
  organization	= {Siemens AG, Vienna, Austria},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baumann93c,
  author	= {Baumann, T. and Strasser, H. and Landrichter, H. },
  title		= {Short-term load forecasting methods in comparison:
		  {K}ohonen learning, backpropagation learning, multiple
		  regression analysis and {K}alman filters},
  booktitle	= {PSCC. Proceedings of the Eleventh Power Systems
		  Computation Conference},
  year		= {1993},
  volume	= {1},
  pages		= {445--51},
  organization	= {Siemens AG, Wien, Austria},
  publisher	= {Power Syst. Comput. Conference},
  address	= {Zurich, Switzerland},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bayer93a,
  author	= {Harald Bayer},
  title		= {{SUSOM} 'SUpervised' Self-Organizing Maps},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {620},
  publisher	= {Springer},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baykal91a,
  author	= {N. Baykal and N. Yalabik and A. H. Goktogan},
  title		= {Character recognition using {K}ohonen's feature map},
  booktitle	= {Computer and Information Sciences VI. Proc. 1991 Int.
		  Symposium},
  year		= {1991},
  editor	= {M. Baray and B. Ozguc},
  volume	= {II},
  pages		= {923--932},
  publisher	= {Elsevier},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@Article{	  baykal92a,
  author	= {Baykal, N. and Yalabik, N. },
  title		= {Object orientation detection and character recognition
		  using optimal feedforward network and {K}ohonen's feature
		  map},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1992},
  volume	= {1709},
  number	= {pt. 1},
  pages		= {292--303},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  baykal98a,
  author	= {Baykal, N. and Erkmen, A. M.},
  title		= {An elastic potential field approach to self-organizing
		  map},
  booktitle	= {IEEE ICIPS'98. Proceedings of the Second IEEE
		  International Conference on Intelligent Processing Systems.
		  IEEE, Piscataway, NJ, USA},
  year		= {1998},
  volume	= {},
  pages		= {555--9},
  abstract	= {This paper proposes an escape methodology to the local
		  minima problem of self-organizing feature maps generated in
		  the overlapping regions which are equidistant to the
		  corresponding winners. The method is based on the
		  artificial potential field concept. Our approach introduces
		  an excitation term, which increases the convergence speed
		  and efficiency of the algorithm while increasing the
		  probability of escaping from local minima. We also
		  associate a learning set which specifies variable
		  attractive and repulsive field of output neurons. Results
		  indicate that accuracy percentile of this method are higher
		  than the original algorithm while it has the ability to
		  escape from local minima and distribute learning in
		  decoupled regions of the winning frequency map.},
  dbinsdate	= {2002/1}
}

@Article{	  baykal99a,
  author	= {Baykal, Nazife and Erkmen, Aydan M.},
  title		= {Extended Self Organizing Feature Map: A tagged potential
		  field approach},
  journal	= {Neural Processing Letters},
  year		= {1999},
  number	= {1},
  volume	= {10},
  pages		= {57--72},
  abstract	= {This paper proposes an escape methodology to the local
		  minima problem of self organizing feature maps generated in
		  the overlapping regions which are equidistant to the
		  corresponding winners. Two new versions of the Self
		  Organizing Feature Map are derived equipped with such a
		  methodology. The first approach introduces an excitation
		  term, which increases the convergence speed and efficiency
		  of the algorithm, while increasing the probability of
		  escaping from local minima. In the second approach, we
		  associate a learning set which specifies the attractive and
		  repulsive field of output neurons. Results indicate that
		  accuracy percentile of the new methods are higher than the
		  original algorithm while they have the ability to escape
		  from local minima.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  beard89a,
  author	= {R. A. Beard and K. S. Rattan},
  title		= {A neural network system for robot vision},
  booktitle	= {Proc. NAECON 1989, IEEE 1989 National Aerospace and
		  Electronics Conference },
  year		= {1989},
  volume	= {IV},
  pages		= {1920--1921},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  beauge93a,
  author	= {Beauge, L. and Durand, S. and Alexandre, F. },
  title		= {Plausible \mbox{self-organizing} maps for speech
		  recognition},
  booktitle	= {Artificial Neural Nets and Genetic Algorithms. Proceedings
		  of the International Conference},
  year		= {1993},
  editor	= {Albrecht, R. F. and Reeves, C. R. and Steele, N. C. },
  pages		= {221--6},
  organization	= {CRIN-CNRS/INRIA Lorraine, Vandoeuvre-les-Nancy, France},
  publisher	= {Springer-Verlag},
  address	= {Berlin, Germany},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bebis91a,
  author	= {G. N. Bebis and G. M. Papadourakis},
  title		= {Model-based object recognition using artificial neural
		  networks},
  booktitle	= {Artificial Neural Networks},
  year		= {1991},
  editor	= {T. Kohonen and K. M{\"{a}}kisara and O. Simula and J.
		  Kangas},
  volume	= {II},
  pages		= {1111--1115},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@Article{	  bebis92a,
  author	= {G. N. Bebis and G. M. Papadourakis},
  title		= {Object recognition using invariant object boundary
		  representations and neural network models},
  journal	= {Pattern Recognition},
  year		= {1992},
  volume	= {25},
  number	= {1},
  pages		= {25--44},
  month		= {January},
  x		= {Several approaches for classifying two-dimensional objects
		  which are based on the use of both invariant boundary
		  transformations and artificial neural networks (ANNs) were
		  implemented and compared. . . . In particular, the
		  multilayer ANN trained with the prediction backpropagation
		  rule and the Kohonen ANN were utilized. },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bebis95a,
  author	= {Bebis, G. and Georgiopoulos, M. and {da Vitoria Lobo}, N.
		  },
  title		= {Learning geometric hashing functions for model-based
		  object recognition},
  booktitle	= {Proceedings of the Fifth International Conference on
		  Computer Vision},
  year		= {1995},
  pages		= {543--8},
  organization	= {Dept. of Electr. \& Comput. Eng. , Univ. of Central
		  Florida, Orlando, FL, USA},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  bebis98a,
  author	= {George Bebis and Michael Georgiopoulos and Niels {da
		  Vitoria Lobo}},
  title		= {Using Self-Organizing Maps to Learn Geometric Hash
		  Functions for Model-Based Object Recognition},
  journal	= {IEEE Transactions on Neural Networks},
  year		= 1998,
  volume	= 9,
  pages		= {560--570},
  abstract	= {A new approach to address the nonuniform distribution of
		  invariants over the hash space during geometric hashing and
		  methods which have emerged from it is proposed. The
		  approach, which is based on an elastic hash table, involves
		  the distribution of the hash bins over the invariants. The
		  key idea is to associate the hash bins with the output
		  nodes of a self-organizing feature map (SOFM) neural
		  network which is trained using the invariants as training
		  examples. In this way, the location of a hash bin in the
		  space of invariants is determined by the weight vector of
		  the node associated with the hash bin. The advantage of
		  this approach is that it is a process that adapts to the
		  invariants through learning.},
  dbinsdate	= {oldtimer}
}

@Article{	  becanovic00a,
  author	= {Becanovic, Vlatko},
  title		= {Image object classification using saccadic search,
		  spatio-temporal pattern encoding and self-organization},
  journal	= {Pattern Recognition Letters},
  year		= {2000},
  number	= {3},
  volume	= {21},
  pages		= {253--263},
  abstract	= {A method for extracting features from photographic images
		  is investigated. The input image is through a saccadic
		  search algorithm divided into a set of sub-images,
		  segmented and coded by a spatio-temporal encoding engine.
		  The input image is thus represented by a set of
		  characteristic pattern signatures, well suited for
		  classification by an unsupervised neural network. A
		  strategy using multiple self-organizing feature maps (SOM)
		  in a hierarchical manner is used. With this approach, using
		  a certain degree of user selection, a database of
		  sub-images is grouped according to similarities in
		  signature space. In English 22 Refs.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  beckenkamp00a,
  author	= {Fábio Ghignatti Beckenkamp and Wolfgang Pree},
  title		= {Building Neural Network components},
  booktitle	= {Proceeding of the ICSC Symposia on Neural Computation
		  (NC'2000) May 23-26, 2000 in Berlin, Germany},
  editor	= {H. Bothe and R. Rojas},
  year		= {2000},
  organization	= {C. Doppler Lab for Software Research, University of
		  Constance},
  publisher	= {ICSC Academic Press},
  dbinsdate	= {oldtimer}
}

@Article{	  beckers97a,
  author	= {M. L. M. Beckers and W. J. Melssen and L. M. C. Buydens},
  title		= {A \mbox{self-organizing} feature map for clustering
		  nucleic acids. Application to a data matrix containing
		  A-DNA and B-DNA dinucleotides},
  journal	= {Computers \& Chemistry},
  year		= {1997},
  volume	= {21},
  number	= {6},
  pages		= {377--90},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  becks92a,
  author	= {K. -H. Becks and J. Dahm and F. Seidel},
  title		= {Analysing particle jets with artificial neural networks},
  booktitle	= {Industrial and Engineering Applications of Artificial
		  Intelligence and Expert Systems. 5th International
		  Conference, IEA/AIE-92},
  year		= {1992},
  editor	= {F. Belli and F. J. Radermacher},
  pages		= {109--112},
  organization	= {Univ. Paderborn; Southwest Texas State Univ; et al},
  publisher	= {Springer},
  address	= {Berlin, Heidelberg},
  dbinsdate	= {oldtimer}
}

@Article{	  bednar00a,
  author	= {Bednar, J. A. and Miikkulainen, R.},
  title		= {Tilt aftereffects in a self-organizing model of the
		  primary visual cortex},
  journal	= {NEURAL COMPUTATION},
  year		= {2000},
  volume	= {12},
  number	= {7},
  month		= {JUL},
  pages		= {1721--1740},
  abstract	= {RF-LISSOM, a self-organizing model of laterally connected
		  orientation maps in the primary visual cortex, was used to
		  study the psychological phenomenon known as the tilt
		  aftereffect. The same self-organizing processes that are
		  responsible for the long-term development of the map are
		  shown to result in tilt aftereffects over short timescales
		  in the adult. The model permits simultaneous observation of
		  large numbers of neurons and connections, making it
		  possible to relate high-level phenomena to low-level
		  events, which is difficult to do experimentally. The
		  results give detailed computational support for the
		  long-standing conjecture that the direct tilt aftereffect
		  arises from adaptive lateral interactions between feature
		  detectors. They also make a new prediction that the
		  indirect effect results from the normalization of synaptic
		  efficacies during this process. The model thus provides a
		  unified computational explanation of self-organization and
		  both the direct and indirect tilt aftereffect in the
		  primary visual cortex.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bednar00b,
  author	= {Bednar, J. A. and Miikkulainen, R.},
  title		= {Self-organization of innate face preferences: could
		  genetics be expressed through learning?},
  booktitle	= {Proceedings Seventeenth National Conference on Artificial
		  Intelligence (AAAI-2000). Twelfth Innovative Applications
		  of Artificial Intelligence Conference (IAAI-2000). AAAI
		  Press, Menlo Park, CA, USA},
  year		= {2000},
  volume	= {},
  pages		= {117--22},
  abstract	= {Self-organizing models develop realistic cortical
		  structures when given approximations of the visual
		  environment as input, and are an effective way to model the
		  development of face recognition abilities. However,
		  environment-driven self-organization alone cannot account
		  for the fact that newborn human infants will preferentially
		  attend to face-like stimuli even immediately after birth.
		  Previously it has been proposed that internally generated
		  input patterns, such as those found in the developing
		  retina and in PGO waves during REM sleep, may have the same
		  effect on self-organization as does the external
		  environment. Internal pattern generators constitute an
		  efficient way to specify, develop, and maintain
		  functionally appropriate perceptual organization. They may
		  help express complex structures from minimal genetic
		  information, and retain this genetic structure within a
		  highly plastic system. Simulations with the CRF-LISSOM
		  model show that such preorganization can account for
		  newborn face preferences, providing a computational
		  framework for examining how genetic influences interact
		  with experience to construct a complex system.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bednar97a,
  author	= {J. A. Bednar and R. Miikkulainen},
  title		= {A Neural Network Model of Visual Tilt Aftereffects},
  booktitle	= {Proceedings of the 19th Annual Meeting of the Cognitive
		  Science Society (COGSCI-97)},
  year		= {1997},
  publisher	= {Erlbaum},
  address	= {Hillsdale, NJ},
  pages		= {37--42},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bednar98a,
  author	= {J. Bednar and R. Miikkulainen},
  title		= {Pattern-Generator-Driven Development in Self-Organizing
		  Models},
  booktitle	= {Computational Neuroscience: Trends in Research},
  publisher	= {Plenum},
  year		= {1998},
  editor	= {J. M. Bower},
  address	= {New York},
  pages		= {317--323},
  dbinsdate	= {oldtimer}
}

@Article{	  beger01a,
  author	= {Beger, R. D. and Wilkes, J. G.},
  title		= {Developing C-13 {NMR} quantitative spectrometric
		  data-activity relationship ({QSDAR}) models of steroid
		  binding to the corticosteroid binding globulin},
  journal	= {JOURNAL OF COMPUTER-AIDED MOLECULAR DESIGN},
  year		= {2001},
  volume	= {15},
  number	= {7},
  month		= {JUL},
  pages		= {659--669},
  abstract	= {We have developed four quantitative spectrometric
		  data-activity relationship (QSDAR) models for 30 steroids
		  binding to corticosteroid binding globulin, based on
		  comparative spectral analysis (CoSA) of simulated C-13
		  nuclear magnetic resonance (NMR) data. A QSDAR model based
		  on 3 spectral bins had an explained variance (r(2)) of 0.80
		  and a cross-validated variance (q(2)) of 0.78. Another
		  QSDAR model using the 3 atoms from the comparative
		  structurally assigned spectral analysis (CoSASA) of
		  simulated C-13 NMR on a steroid backbone template gave an
		  explained variance (r(2)) of 0.80 and a cross-validated
		  variance (q(2)) of 0.73. Positions 3 and 14 from the
		  steroid backbone template have correlations with the
		  relative binding activity to corticosteroid binding
		  globulin that are greater than 0.52. The explained
		  correlation and cross-validated correlation of these QSDAR
		  models are as good as previously published quantitative
		  structure-activity relationship (QSAR), self-organizing map
		  (SOM) and electrotopological state (E- state) models. One
		  reason that the cross-validated variance of QSDAR models
		  were as good as the other models is that simulated C-13 NMR
		  spectral data are more accurate than the errors introduced
		  by the assumptions and approximations used in calculated
		  electrostatic potentials, E-states, HE-states, and the
		  molecular alignment process of QSAR modeling. The QSDAR
		  models developed provide a rapid, simple way to predict the
		  binding activity of a steroid to corticosteroid binding
		  globulin.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  behera95a,
  author	= {Behera, L. and Gopal, M. and Chaudhury, S. },
  title		= {Self-organizing neural networks for learning inverse
		  dynamics of robot manipulator},
  booktitle	= {1995 IEEE/IAS International Conference on Industrial
		  Automation and Control (I A \& C'95)},
  year		= {1995},
  pages		= {457--60},
  organization	= {Dept. of Electr. Eng. , Indian Inst. of Technol. , New
		  Delhi, India},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  behera98a,
  author	= {L. Behera and S. Chaudhury and M. Gopal},
  title		= {Application of Self Organizing Neural Networks in Robot
		  Tracking Control},
  journal	= {IEE Proceedings---Control Theory and Applications},
  volume	= {145},
  pages		= {135--140},
  year		= {1998},
  dbinsdate	= {oldtimer}
}

@Article{	  behera99a,
  author	= {Behera, Laxmidhar and Krishna, GPrem},
  title		= {Neural controller based binocular vision system for object
		  tracking},
  journal	= {IETE Journal of Research},
  year		= {1999},
  number	= {1},
  volume	= {45},
  pages		= {63--71},
  abstract	= {ln this paper, we present a learning control approach to
		  the problem of visual tracking using an active binocular
		  vision system. The simulated vision system used here is
		  composed of two cameras which have freedom of pan and are
		  mounted on a plate that can tilt. A self organizing map is
		  trained to learn the mapping between visual feedback as
		  received from the simulated CCD cameras and the incremental
		  values in the orientation angles of the vision system to
		  fixate continuously on a mobile target. A direct
		  implication of this is that the object should be centered
		  at the vergence point of the cameras. This is accomplished
		  by ensuring minimum disparity of the target vector from the
		  center of the images of both the cameras. The learning
		  algorithm is presented and the simulation study shows an
		  average focussing error of 0.15 pixels after 1200
		  iterations for a small, defined joint angle space of the
		  vision system. Thus, a significant contribution of this
		  work is the application of a self organizing map in dynamic
		  object tracking.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  behme93a,
  author	= {Holger Behme and Wolf Dieter Brandt and Hans Werner
		  Strube},
  title		= {Speech Recognition by Hierarchical Segment
		  Classification},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {416--419},
  publisher	= {Springer},
  address	= {London, UK},
  abstract	= {A neural network for speech processing is presented. Its
		  complex architecture, incorporating self-organizing feature
		  maps [1], allows the construction of a hierarchy of layers,
		  where each layer operates on a larger time scale and deals
		  with higher units of speech, like phonemes, syllables, word
		  parts and so on. Tasks the network has to deal with include
		  representation of speech, segmentation of the speech signal
		  and classifying segments.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  behme93b,
  author	= {Holger Behme and Wolf Dieter Brandt and Hans Werner
		  Strube},
  title		= {Speech Processing by Hierarchical Segment Classification},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {I},
  pages		= {279--282},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  beilharz95a,
  author	= {Beilharz, J. and Ropke, K. and Filbert, D. },
  title		= {Statistical and neural concepts of unsupervised classifier
		  design for motor diagnosis},
  journal	= {Automatisierungstechnik},
  year		= {1995},
  volume	= {43},
  number	= {1},
  pages		= {46--53},
  month		= {Jan},
  dbinsdate	= {oldtimer}
}

@Article{	  belic97a,
  author	= {I. Belic and L. Gyergyek},
  title		= {Neural network methodologies for mass spectra
		  recognition},
  journal	= {Vacuum},
  year		= {1997},
  volume	= {48},
  number	= {7--9},
  pages		= {633--7},
  note		= {(5th European Vacuum Conference, EVC-5 Conf. Date: 23--27
		  Sept. 1996 Conf. Loc: Salamanca, Spain)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bellando96a,
  author	= {J. Bellando and R. Kothari},
  title		= {On image correspondence using topology preserving
		  mappings},
  booktitle	= {ICNN 96. The 1996 IEEE International Conference on Neural
		  Networks},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {3},
  address	= {New York, NY, USA},
  pages		= {1784--9},
  abstract	= {A computational approach for establishing correspondence
		  between two image views is presented. We show that a
		  self-organizing feature map trained with tokens (features)
		  from the first frame and subsequently with tokens from the
		  second frame (without re-initialization) is capable of
		  indicating the underlying transformation which results in
		  the second frame. The reliance of the self-organizing
		  feature map on the underlying probability density of the
		  features makes the proposed approach insensitive to missing
		  tokens. Simulation results under three different observer
		  movements (panning, zooming, and rotation) are presented to
		  illustrate the proposed method.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bellido93a,
  author	= {I. Bellido and E. Fiesler},
  title		= {Do Backpropagation Trained Neural Networks have Normal
		  Weight Distributions},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {772--775},
  publisher	= {Springer},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@Article{	  bellotti95a,
  author	= {Bellotti, R. and Castellano, M. and {De Marzo}, C. and
		  Satalino, G. },
  title		= {Signal/background classification in a cosmic ray space
		  experiment by a modular neural system},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1995},
  volume	= {2492},
  number	= {pt. 2},
  pages		= {1153--61},
  publisher	= {SPIE-Int. Soc. Opt. Eng},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@Article{	  benaim93a,
  author	= {Michel Benaim},
  title		= {The 'Off Line Learning Approximation' in Continuous Time
		  Neural Networks: An Adiabatic Theorem},
  journal	= {Neural Networks},
  year		= {1993},
  volume	= {6},
  number	= {5},
  pages		= {655--665},
  dbinsdate	= {oldtimer}
}

@InCollection{	  benaim97a,
  author	= {M. Benaim and J. -C. Fort and G. Pages},
  title		= {Almost sure convergence of the
		  \mbox{\mbox{one-dimensional}} {K}ohonen algorithm},
  booktitle	= {5th European Symposium on Artificial Neural Networks ESANN
		  '97. Proceedings},
  publisher	= {D facto},
  year		= {1997},
  editor	= {M. Verleysen},
  address	= {Brussels, Belgium},
  pages		= {193--8},
  dbinsdate	= {oldtimer}
}

@Article{	  benaim98a,
  author	= {M. Benaim and J. -C. Fort and G. Pages},
  title		= {Convergence of the \mbox{\mbox{one-dimensional}} {K}ohonen
		  algorithm},
  journal	= {Advances in Applied Probability},
  year		= {1998},
  volume	= {30},
  number	= {3},
  pages		= {850--69},
  dbinsdate	= {oldtimer}
}

@InCollection{	  benaki94a,
  author	= {A. Benaki and B. Gatos and I. Karamani and D. Karras and
		  S. Perantonis and N. Vassilas and N. Gaitanis},
  title		= {A robot hand-eye coordination system for {3-D} object
		  recognition using novel neural networks trained with
		  multiview moments},
  booktitle	= {Proceedings EURISCON `94. European Robotics and
		  Intelligent Systems Conference},
  publisher	= {Univ. Bristol},
  year		= {1994},
  volume	= {3},
  address	= {Bristol, UK},
  pages		= {1692--701},
  dbinsdate	= {oldtimer}
}

@Book{		  bengtsson96a,
  author	= {Bengtsson, M.},
  title		= {Hierarchical Clustering in a Mean Field Theory. Scientific
		  rept.},
  year		= {1996},
  abstract	= {A hierarchical clustering model, which is built on a Potts
		  glass in a mean field approximation, is developed. It
		  gives, we believe, the optimal clustering solution to
		  randomly positioned Gaussian clusters of identical size and
		  standard deviation. The phase transition temperatures are
		  estimated in advance, facilitating a completely parameter
		  free model, and thereby avoiding parameter tuning that
		  plagues other self-organizing models. Explored numerically,
		  the model performs well when clustering problems are scaled
		  up. Furthermore, it is demonstrated that standard vector
		  quantization and the Kohonen self-organizing map are unable
		  to solve these clustering tasks in general.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  benitez-diaz94a,
  author	= {Benitez-Diaz, D. and Carrabina, J. and Gonzalez-Rodriguez,
		  M. },
  title		= {Neural-like network model for color images analysis
		  systems},
  booktitle	= {1994 IEEE International Conference on Neural Networks.
		  IEEE World Congress on Computational Intelligence},
  year		= {1994},
  volume	= {3},
  pages		= {1415--20},
  organization	= {Departamento de Inf. y Sistemas, Campus Univ. de Tafira,
		  Las Palmas, Spain},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  benitez-diaz95a,
  author	= {D. Benitez-Diaz and J. Garcia-Quesada},
  title		= {Learning algorithm with {G}aussian membership function for
		  fuzzy {RBF} neural networks},
  booktitle	= {From Natural to Artificial Neural Computation.
		  International Workshop on Artificial Neural Networks.
		  Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1995},
  editor	= {J. Mira and F. Sandoval},
  address	= {Berlin, Germany},
  pages		= {527--34},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  benitez00a,
  author	= {Benitez, Cesar and Lander, Daniel Kvedaras and Ramirez,
		  Jose},
  title		= {Reduction of dimensionality for perceptual clustering},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2000},
  editor	= {},
  volume	= {5},
  pages		= {148--151},
  organization	= {Universidad Simon Bolivar},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {Multidimensionality is one of the problems to be solved
		  for a robust methodology in order to be capable of
		  resolving simple and realistic problems. This work
		  establishes a complete methodology based on self-organized
		  maps (SOM) and the expectation-maximization (EM) algorithm
		  that finds an abstract probability function, which is a mix
		  of local experts. An application of this methodology is
		  presented as a case study, where the problem is robot
		  navigation in noisy environments. Readings from seven robot
		  sonars were taken as input for the system, mapped into a
		  two dimension space and grouped into abstract observations,
		  in order to make recognition of navigation space
		  environment dependant and accurate. The goal is to build
		  the capability of predicting observations and of
		  recognizing abstractions that were defined over the data
		  itself.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  benitez00b,
  author	= {Benitez, L. M. and Ferreira, A. and Iparraguirre, D. I.
		  P.},
  title		= {Chaotic time series prediction using the Kohonen
		  algorithm},
  booktitle	= {8th European Symposium on Artificial Neural Networks.
		  ESANN"2000. Proceedings. D-Facto, Brussels, Belgium},
  year		= {2000},
  volume	= {},
  pages		= {347--52},
  abstract	= {Deterministic nonlinear prediction is a powerful technique
		  for the analysis and prediction of time series generated by
		  nonlinear dynamical systems. In this paper, the use of a
		  Kohonen network as a component of one deterministic
		  nonlinear prediction algorithm is suggested. In order to
		  evaluate the performance of the proposed algorithm, it was
		  applied to the prediction of time series generated by two
		  well-known chaotic dynamical systems, and the results were
		  compared with those obtained using the modified method of
		  analogues with the same time series. The generated time
		  series were corrupted by superimposed observational noise.
		  The experimental results have shown that the Kohonen
		  network can learn the neighborhood relations present in the
		  reconstructed attractor of the time series and that good
		  predictions can also be obtained with the proposed
		  algorithm.},
  dbinsdate	= {2002/1}
}

@Article{	  benitez00c,
  author	= {Benitez, M. C. and Rubio, A. and Garcia, P. and {de la
		  Torre}, A.},
  title		= {Different confidence measures for word verification in
		  speech recognition},
  journal	= {Speech Communication},
  year		= {2000},
  volume	= {32},
  number	= {1},
  month		= {Sep},
  pages		= {79--94},
  organization	= {Universidad de Granada},
  publisher	= {Elsevier Science Publishers B.V.},
  address	= {Amsterdam},
  abstract	= {Recent research in Automatic Speech Recognition (ASR)
		  technologies has shown the key-word spotting (KWS) systems
		  as one of the most interesting options for accessing
		  information using speech. KWS systems can accept
		  spontaneous speech, which allows potential users to ask for
		  information without learning complex protocols for the
		  human-machine communication. One of the most relevant
		  aspects in KWS systems is the verification of key-word
		  candidates. Utterances detected as key-words could be
		  either `false alarms' (non-key-words or incorrectly
		  recognized key-words) or `correct key-words'. The use of
		  confidence measurements allows (by additional processing of
		  the spoken sentence) the verification of the candidates and
		  the decision as to whether each utterance must be accepted
		  as a correctly recognized key-word or rejected as a false
		  alarm. In this work we propose a novel method for
		  verification in those KWS systems based on phone models.
		  Under our new approach, a phonematic speech recognizer
		  decodes the spoken sentence in parallel with the KWS
		  recognizer. The first one produces a phone string as output
		  while the second one generates a key-word/filler-model
		  string. By aligning both strings, a set of characteristics
		  is extracted which are used to verify the putatives
		  key-word. For that we have built two classifiers; in the
		  first one the euclidean metric is modified and adapted in a
		  local and iterative way in order to give greater importance
		  to the most discriminate directions between the classes.
		  The second is a vector quantizer which was trained using
		  adaptative technique learning. We have applied the proposed
		  method to several KWS tasks. Experimental results presented
		  in this paper show that the proposed verification method
		  improves the performance of the KWS systems by reducing the
		  false alarm rate without a significant increase in the
		  rejection of correctly detected keywords.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  benitez_rochel00a,
  author	= {{Benitez Rochel}, R. and {Trella Lopez}, M. and {Conejo
		  Mu\~{n}oz}, R.},
  title		= {Neural networks applied to Item Response Theory},
  booktitle	= {Proceeding of the ICSC Symposia on Neural Computation
		  (NC'2000) May 23-26, 2000 in Berlin, Germany},
  editor	= {H. Bothe and R. Rojas},
  year		= {2000},
  organization	= {Departamento de Lenguajes y Ciencias de la Computación
		  Universidad de Málaga},
  publisher	= {ICSC Academic Press},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bennani90a,
  author	= {Y. Bennani and F. Fogelman-Souli{\'{e}} and P. Gallinari},
  title		= {A connectionist approach for automatic speaker
		  identification},
  booktitle	= {Proc. ICASSP-90, International Conference on Acoustics,
		  Speech and Signal Processing},
  year		= {1990},
  volume	= {I},
  pages		= {265--268},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bennani90b,
  author	= {Y. Bennani and N. Chaourar and P. Gallinari and A.
		  Mellouk},
  title		= {Comparing neural net models on speech recognition tasks},
  booktitle	= {Proc. Neuro-N\^{i}mes '90, Third Int. Workshop. Neural
		  Networks and Their Applications},
  year		= {1990},
  pages		= {455--467},
  organization	= {ARC; JSAI; SEE},
  publisher	= {EC2},
  address	= {Nanterre, France},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bennani90c,
  author	= {Y. Bennani and F. Fogelman-Souli{\'{e}} and P. Gallinari},
  title		= {Text-dependent speaker identification using learning
		  vector quantization},
  booktitle	= {Proc. INNC'90, Int. Neural Network Conference },
  year		= {1990},
  volume	= {II},
  pages		= {1087--1090},
  organization	= {Thomsom; SUN; British Computer Society ; et al},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  annote	= {page numbers may be incorrect},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bennani91a,
  author	= {Y. Bennani and N. Chaourar and P. Gallinari and A.
		  Mellouk},
  title		= {Validation of neural net architectures on speech
		  recognition tasks},
  booktitle	= {Proc. ICASSP-91, International Conference on Acoustics,
		  Speech and Signal Processing },
  year		= {1991},
  volume	= {I},
  pages		= {97--100},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bennani99a,
  author	= {Bennani, Y.},
  title		= {Adaptive weighting of pattern features during learning},
  booktitle	= {IJCNN'99. International Joint Conference on Neural
		  Networks. Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {5},
  pages		= {3008--13},
  abstract	= {Irrelevant or redundant features may have negative effects
		  on classification algorithms. The designer of a
		  classification algorithm typically require a few very
		  significant features characterising the class membership of
		  the patterns. The discriminatory information is encoded in
		  a very complex manner and features which are the most
		  important for pattern classification may not be apparent.
		  One way to address this problem is the use of feature
		  weighting procedure as a data preprocessing step. In this
		  paper we propose a two-step algorithm as an extension of
		  the learning vector quantization algorithm (LVQ). This
		  approach is based on weighting features depending on their
		  contribution to discrimination. Adapting weighting
		  coefficients and codewords is done simultaneously by using
		  a new global learning algorithm named omega LVQ2.
		  Experiments are undertaken on a synthetic problem and on
		  real problems in speech and speaker recognition domain, to
		  show significant improvement over the standard learning
		  algorithm.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  benson00a,
  author	= {Benson, Karl},
  title		= {Evolving Finite State Machines with embedded Genetic
		  Programming for Automatic Target Detection},
  booktitle	= {Proceedings of the IEEE Conference on Evolutionary
		  Computation, ICEC},
  year		= {2000},
  editor	= {},
  volume	= {2},
  pages		= {1543--1549},
  organization	= {Defence Evaluation and Research Agency Malvern},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {This paper presents a model comprising Finite State
		  Machines (FSMs) with embedded Genetic Programs (GPs) which
		  co-evolve to perform the task of Automatic Target Detection
		  (ATD). The fusion of a FSM and GPs allows for a control
		  structure (main program), the FSM, and sub-programs, the
		  GPs, to co-evolve in a symbiotic relationship. The GP
		  outputs along with the FSM state transition levels are used
		  to construct confidence intervals that enable each pixel
		  within the image to be classified as either target or
		  non-target, or to cause a state transition to take place
		  and further analysis of the pixel to be performed. The
		  algorithms produced using this method consist of nominally
		  four GPs, with a typical node cardinality of less than ten,
		  that are executed in an order dictated by the FSM. The
		  results of the experimentation performed are compared to
		  those obtained in two independent studies of the same
		  problem using Kohonen Neural Networks and a two stage
		  Genetic Programming strategy.},
  dbinsdate	= {2002/1}
}

@Article{	  benson00b,
  author	= {Benson, M. W. and Hu, J.},
  title		= {Asynchronous self-organizing maps},
  journal	= {IEEE TRANSACTIONS ON NEURAL NETWORKS},
  year		= {2000},
  volume	= {11},
  number	= {6},
  month		= {NOV},
  pages		= {1315--1322},
  abstract	= {A recently defined energy function which leads to a self-
		  organizing map is used as a foundation for an asynchronous
		  neural-network algorithm, We generalize the existing
		  stochastic gradient approach to an asynchronous parallel
		  stochastic gradient method for generating a topological map
		  on a distributed computer system (MIMD). A convergence
		  proof is presented and simulation results on a set of
		  problems are included. A practical problem using the energy
		  function approach is that a summation over the entire
		  network is required during the computation of updates.
		  Using simulations we demonstrate effective algorithms that
		  use efficient sampling for the approximation of these
		  sums.},
  dbinsdate	= {2002/1}
}

@TechReport{	  beppu90a,
  author	= {T. Beppu and M. Sase and Y. Kosugi},
  title		= {Self-Organizing Feature Map using Classified Neural
		  Units},
  institution	= {The Inst. of Electronics, Information and Communication
		  Engineers},
  year		= {1990},
  number	= {PRU90--96},
  address	= {Tottori University, Koyama, Japan},
  note		= {(in Japanese)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  berenji93a,
  author	= {Hamid R. Berenji},
  title		= {Neural Networks for Fuzzy Logic Inference},
  booktitle	= {Proc. International Conference on Fuzzy Systems},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1993},
  pages		= {1395},
  abstract	= {The non-adaptive behavior of current fuzzy inference
		  systems can significantly be improved by using artificial
		  neural networks. Recent interests in combining fuzzy logic
		  and neural networks have led to the development of new
		  algorithms that provide adaptive behavior while maintaining
		  the strong knowledge representation characteristic of fuzzy
		  systems. Many perspectives on the impact of neural networks
		  on fuzzy logic inference exist, however, in this plenary
		  talk, we focus on four important categories: (a)- pattern
		  recognition, (b)- formation of fuzzy rules from data, (c)-
		  refinement of fuzzy membership functions, and (d)-
		  implementation of fuzzy inference systems using neural net
		  works. In pattern recognition area, we review the hybrid
		  neural networks and fuzzy logic methods for cluster
		  analysis in fuzzy sets. The Kohonen's Feature Maps, Fuzzy
		  ART, and Bezdek's Fuzzy-LVQ are considered. The temporal
		  back-propagation algorithm, and radial basis functions can
		  be used to extract rules from data. On the other hand,
		  reinforcement learning, and supervised learning techniques
		  such as backpropagation can be used in fine-tuning the
		  membership functions. Finally, the advantages of using
		  neural networks for implementing fuzzy rules will be
		  discussed and examples of their applications will be
		  demonstrated.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  berger97a,
  author	= {A. Berger and D. P. F. Moller and M. Renter},
  title		= {Detection of sleep with new preprocessing methods for EEG
		  analysing},
  booktitle	= {Computational Intelligence Theory and Applications.
		  International Conference, 5th Fuzzy Days. Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {B. Reusch},
  address	= {Berlin, Germany},
  pages		= {304--10},
  dbinsdate	= {oldtimer}
}

@Article{	  bermejo00a,
  author	= {Bermejo, Sergio and Cabestany, Joan},
  title		= {Batch learning vector quantization algorithm for nearest
		  neighbour classification},
  journal	= {Neural Processing Letters},
  year		= {2000},
  volume	= {11},
  number	= {3},
  month		= {Jun},
  pages		= {173--184},
  organization	= {Universitat Politecnica de Catalunya (UPC)},
  publisher	= {Kluwer Academic Publishers},
  address	= {Dordrecht},
  abstract	= {We introduce a batch learning algorithm to design the set
		  of prototypes of 1 nearest-neighbour classifiers. Like
		  Kohonen's LVQ algorithms, this procedure tends to perform
		  vector quantization over a probability density function
		  that has zero points at Bayes borders. Although it differs
		  significantly from their online counterparts since: (1) its
		  statistical goal is clearer and better defined; and (2) it
		  converges superlinearly due to its use of the very fast
		  Newton's optimization method. Experiments results using
		  artificial data confirm faster training time and better
		  classification performance than Kohonen's LVQ algorithms.},
  dbinsdate	= {2002/1}
}

@Article{	  bermejo01a,
  author	= {Bermejo, S. and Cabestany, J.},
  title		= {Learning with nearest neighbour classifiers},
  journal	= {Neural Processing Letters},
  year		= {2001},
  volume	= {13},
  number	= {2},
  month		= {April 2001},
  pages		= {159--181},
  organization	= {Department of Electronic Engineering, Univ. Politecnica de
		  Catalunya, C4 building},
  publisher	= {},
  address	= {},
  abstract	= {This paper introduces a learning strategy for designing a
		  set of prototypes for a 1-nearest-neighbour (NN)
		  classifier. In learning phase, we transform the 1-NN
		  classifier into a maximum classifier whose discriminant
		  functions use the nearest models of a mixture. Then the
		  computation of the set of prototypes is viewed as a problem
		  of estimating the centres of a mixture model. However,
		  instead of computing these centres using standard
		  procedures like the EM algorithm, we derive to compute a
		  learning algorithm based on minimising the
		  misclassification accuracy of the 1-NN classifier on the
		  training set. One possible implementation of the learning
		  algorithm is presented. It is based on the online gradient
		  descent method and the use of radial gaussian kernels for
		  the models of the mixture. Experimental results using
		  hand-written NIST databases show the superiority of the
		  proposed method over Kohonen's LVQ algorithms.},
  dbinsdate	= {2002/1}
}

@Article{	  bermejo01b,
  author	= {Bermejo, S. and Cabestany, J.},
  title		= {Finite-sample convergence properties of the {LVQ}1
		  algorithm and the batch {LVQ}1 algorithm},
  journal	= {Neural Processing Letters},
  year		= {2001},
  volume	= {13},
  number	= {2},
  month		= {April 2001},
  pages		= {135--157},
  organization	= {Department of Electronic Engineering, Univ. Politecnica de
		  Catalunya, C4 building},
  publisher	= {},
  address	= {},
  abstract	= {This letter addresses the asymptotic convergence of
		  Kohonen's LVQ1 algorithm when the number of training
		  samples are finite with an analysis that uses the dynamical
		  systems and optimisation theories. It establishes the
		  sufficient conditions to ensure the convergence of LVQ1
		  near a minimum of its cost function for constant step sizes
		  and cyclic sampling. It also proposes a batch version of
		  LVQ1 based on the very fast Newton optimisation method that
		  cancels the dependence of the on-line version on the order
		  of supplied training samples.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bermejo98a,
  author	= {Bermejo, S. and Cabestany, J. and Payeras, M.},
  title		= {A new dynamic {LVQ} -based classifier and its application
		  to handwritten character recognition},
  booktitle	= {6th European Symposium on Artificial Neural Networks.
		  ESANN'98. Proceedings},
  publisher	= {D-Facto},
  address	= {Brussels, Belgium},
  year		= {1998},
  volume	= {},
  pages		= {203--8},
  abstract	= {In this paper we present a new dynamic learning strategy
		  in LVQ's framework that consists of a growing process and a
		  posteriori pruning process. This schema allows building a
		  family of dynamic LVQ learning systems. To evaluate this
		  proposal, we have employed LVQ3 as a core of the learning
		  system. We have done an empirical analysis of this system
		  (DLVQ3) in order to characterize its parameters and to
		  compare it with nondynamical LVQ algorithms. Finally we
		  present results on applying PCA/sub 64/+DLVQ3 to upper and
		  lower handwritten character recognition, obtaining on
		  average test classification error (with no rejection) 7.22%
		  and 15.275% respectively in front of 9.16% and 17.01%
		  achieved with PCA/sub 64/+nondynamic LVQx's of similar
		  size, 14.7% and 23.1% with PCA/sub 64/+PNN and 10.1% and
		  20.3% with PCA/sub 128/+MLP.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bermejo99a,
  author	= {Bermejo, S. and Cabestany, J.},
  title		= {Online gradient learning algorithms for k-nearest neighbor
		  classifiers},
  booktitle	= {Foundations and Tools for Neural Modeling. International
		  Work-Conference on Artificial and Natural Neural Networks,
		  IWANN'99. Proceedings, (Lecture Notes in Computer Science
		  Vol.1606)},
  publisher	= {Springer-Verlag},
  address	= {Berlin, Germany},
  year		= {1999},
  volume	= {1},
  pages		= {546--55},
  abstract	= {We present two online gradient learning algorithms to
		  design condensed k-nearest neighbor (NN) classifiers. The
		  goal of these learning procedures is to minimize a measure
		  of performance closely related to the expected
		  misclassification rate of the k-NN classifier. One possible
		  implementation of the algorithm is given. Convergence
		  properties are analyzed and connections with other works
		  are established. We compare these learning procedures with
		  Kohonen's LVQ algorithms (1996) and k-NN classification
		  using the handwritten NIST databases. Experimental results
		  demonstrate the potential of the proposed learning
		  algorithms.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bermudez96a,
  author	= {J. L. Bermudez and A. Piras and M. Rubinstein},
  title		= {Classification of lightning electromagnetic waveforms with
		  a \mbox{self-organizing} {K}ohonen map},
  booktitle	= {Thirteenth International Wroclaw Symposium Electromagnetic
		  Compatibility 1996},
  publisher	= {Inst. Telecommun},
  year		= {1996},
  editor	= {J. M. Janiszewski and W. Moron and W. Sega},
  address	= {Warsaw, Poland},
  pages		= {517--21},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bernard01a,
  author	= {Bernard, S. and Boujemaa, N. and Vitale, D. and Bricot,
		  C.},
  title		= {Fingerprint classification using Kohonen topological map},
  booktitle	= {IEEE International Conference on Image Processing},
  year		= {2001},
  editor	= {},
  volume	= {3},
  pages		= {230--233},
  organization	= {INRIA Rocquencourt},
  publisher	= {},
  address	= {},
  abstract	= {Self Organizing Maps are efficient and usual for dimension
		  reduction and data clustering. In our present work, we
		  propose the use of Kohonen Topologic Map for fingerprint
		  pattern classification. The learning process takes into
		  account the large intra-class diversity and the continuum
		  of fingerprint pattern types. After a brief introduction to
		  fingerprint domain-specific knowledge and the expert
		  approach, we present an original and intuitive description
		  of the algorithm. For a classification based on the global
		  shape of the fingerprint, we adopted a suitable feature
		  space. Indeed we obtained 88% of correct classification on
		  a database composed of 1600 NIST fingerprints.},
  dbinsdate	= {2002/1}
}

@InCollection{	  bernard97a,
  author	= {Gilles Bernard},
  title		= {Experiments on distributional categorization of lexical
		  items with self organizing maps},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {304--309},
  dbinsdate	= {oldtimer}
}

@Article{	  bernard98a,
  author	= {P. Bernard and A. Golbraikh and D. Kireev and J. R.
		  Chretien and N. Rozhkova},
  title		= {Comparison of Chemical Databases Analysis of Molecular
		  Diversity with Self Organizing Maps {SOM} },
  journal	= {Analusis, An International Journal on Analytical
		  Chemistry},
  volume	= {26},
  pages		= {333--341},
  year		= {1998},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  berns93a,
  author	= {Berns, K. and Muller, B. and Dillmann, R. },
  title		= {Dynamic control of a robot leg with \mbox{self-organizing}
		  feature maps},
  booktitle	= {IROS '93. Proceedings of the 1993 IEEE/RSJ International
		  Conference on Intelligent Robots and Systems. Intelligent
		  Robots for Flexibility},
  year		= {1993},
  volume	= {1},
  pages		= {553--60},
  organization	= {Gruppe Interaktive Planuungstechnik, Forschungszentrum
		  Inf. , Karlsruhe, Germany},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  bertin97a,
  author	= {E. Bertin},
  title		= {Self organizing maps and imaging surveys},
  journal	= {Astrophysics and Space Science Library},
  year		= {1997},
  volume	= {210},
  pages		= {221--2},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bertsch87a,
  author	= {H. Bertsch and J. Dengler},
  title		= {Klassifizierung und {S}egmentierung medizinischer {B}ilder
		  mit {H}ilfe der selbstlernenden topologischen {K}arte},
  booktitle	= {9. DAGM-Symp. Mustererkennung},
  year		= {1987},
  editor	= {E. Paulus},
  pages		= {166--170},
  organization	= {Deutche Arbeitsgruppe f{\"{u}}r Mustererkennung},
  publisher	= {Springer},
  address	= {Berlin},
  x		= {tiedostosta leila. bib, viite aika sekava: {Bertsch H,
		  Dengler J (1987) Klassifizierung und Segmentierung
		  medizinischerBilder mit Hilfe der selbstlernenden
		  topologischen Karte,E Paulus (ed. ), 9. DAGM-Symposium
		  Mustererkennung, 166--170,Springer Informatik Fachberichte
		  149, Berlin, Heidelberg. }},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  betke99a,
  author	= {Betke, M. and Kawai, J.},
  title		= {Gaze detection via \mbox{self-organizing} gray-scale
		  units},
  booktitle	= {Proceedings International Workshop on Recognition,
		  Analysis, and Tracking of Faces and Gestures in Real-Time
		  Systems. In Conjunction with ICCV'99},
  year		= {1999},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA, USA},
  volume	= {},
  pages		= {70--6},
  abstract	= {We present a gaze estimation algorithm that detects an eye
		  in a face image and estimates the gaze direction by
		  computing the position of the pupil with respect to the
		  center of the eye. The algorithm is information conserving
		  and based on unsupervised learning. It creates a map of
		  self-organized gray-scale image units that collectively
		  learn to describe the eye outline.},
  dbinsdate	= {oldtimer}
}

@MastersThesis{	  beveridge93a,
  author	= {Martin Beveridge},
  title		= {Using Self Organizing Maps for the Objective Assesment of
		  /s/ Misarticualtions by Patients with Intra-Oral Cancers},
  school	= {University of Edinburgh, Department of Linguistics},
  year		= {1993},
  address	= {Edinburgh, UK},
  dbinsdate	= {oldtimer}
}

@Article{	  bezdek01a,
  author	= {Bezdek, J. C. and Kuncheva, L. I.},
  title		= {Nearest prototype classifier designs: An experimental
		  study},
  journal	= {International Journal of Intelligent Systems},
  year		= {2001},
  volume	= {16},
  number	= {12},
  month		= {December },
  pages		= {1445--1473},
  organization	= {Computer Science Department, University of West Florida},
  publisher	= {},
  address	= {},
  abstract	= {We compare eleven methods for finding prototypes upon
		  which to base the nearest prototype classifier. Four
		  methods for prototype selection are discussed: Wilson +
		  Hart (a condensation + error-editing method), and three
		  types of combinatorial search-random search, genetic
		  algorithm, and tabu search. Seven methods for prototype
		  extraction are discussed: unsupervised vector quantization,
		  supervised learning vector quantization (with and without
		  training counters), decision surface mapping, a fuzzy
		  version of vector quantization, c-means clustering, and
		  bootstrap editing. These eleven methods can be usefully
		  divided two other ways: by whether they employ pre- or
		  postsupervision; and by whether the number of prototypes
		  found is user-defined or "automatic." Generalization error
		  rates of the 11 methods are estimated on two synthetic and
		  two real data sets. Offering the usual disclaimer that
		  these are just a limited set of experiments, we feel
		  confident in asserting that presupervised, extraction
		  methods offer a better chance for success to the casual
		  user than postsupervised, selection schemes. Finally, our
		  calculations do not suggest that methods which find the
		  "best" number of prototypes "automatically" are superior to
		  methods for which the user simply specifies the number of
		  prototypes. },
  dbinsdate	= {2002/1}
}

@Article{	  bezdek90a,
  author	= {J. C. Bezdek},
  title		= {A note on generalized \mbox{self-organizing} network
		  algorithms},
  journal	= {Proc. SPIE---The International Society for Optical
		  Engineering},
  year		= {1990},
  volume	= {1293},
  number	= {pt. 1},
  pages		= {260--267},
  publisher	= {SPIE},
  address	= {Bellingham, WA},
  note		= {Conference paper in journal},
  abstract	= {We identify some similarities and differences between two
		  clustering models, viz., the fuzzy c-means (FCM) and
		  Kohonen self-organizing (KSO) feature map approaches. This
		  leads us to suggest that there is an important unknown
		  relationship between the two methodologies. Consequently,
		  we propose several avenues of research which, if
		  successfully resolved, will strengthen both the FCM and
		  Kohonen models and their utility for applications in
		  clustering and classifier design.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bezdek91a,
  author	= {J. C. Bezdek},
  title		= {Self-Organization and Clustering Algorithms},
  booktitle	= {Proc. 2nd Joint Technology Workshop on Neural Networks and
		  Fuzzy Logic},
  year		= {1991},
  volume	= {I},
  pages		= {143--158},
  x		= {nt9093. bib---AN ACCESSION NUMBER: N91217836XSP},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bezdek92a,
  author	= {J. C. Bezdek and E. C. -K. Tsao and N. R. Pal},
  title		= {Fuzzy {K}ohonen clustering networks},
  booktitle	= {Proc. IEEE International Conference on Fuzzy Systems},
  year		= {1992},
  pages		= {1035--1043},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bezdek92b,
  author	= {James C. Bezdek},
  title		= {Integration and generalization of {LVQ} and c-means
		  clustering},
  booktitle	= {SPIE Vol. 1826, Intelligent Robots and Computer Vision XI:
		  Biological, Neural Net, and 3-D Methods},
  year		= {1992},
  pages		= {280--299},
  volume	= {1826},
  publisher	= {SPIE},
  address	= {Bellingham, WA},
  abstract	= {This paper discusses the relationship between the
		  sequential hard c-means (SHCM), learning vector
		  quantization (LVQ), and fuzzy c-means (FCM) clustering
		  algorithms. LVQ and SHCM suffer from several major
		  problems. For example, they depend heavily on
		  initialization. If the initial values of the cluster
		  centers are outside the convex hull of the input data, such
		  algorithms, even if they terminate, may not produce
		  meaningful results in terms of prototypes for cluster
		  representation. This is due in part to the fact that they
		  update only the winning prototype for every input vector.
		  We also discuss the impact and interaction of these two
		  families with Kohonen's self-organizing feature mapping
		  (SOFM), which is not a clustering method, but which often
		  lends itself to clustering algorithms. Then we present two
		  generalizations of LVQ that are explicitly designed as
		  clustering algorithms: we refer to these algorithms as
		  generalized LVQ = GLVQ; and fuzzy LVQ = FLVQ. Learning
		  rules are derived to optimize an objective function whose
		  goal is to produce 'good clusters'. GLVQ/FLVQ (may) update
		  every node in the clustering net for each input vector. We
		  use Anderson's IRIS data to compare the performance of
		  GLVQ/FLVQ with a standard version of LVQ. Experiments show
		  that the final centroids produced by GLVQ are independent
		  of node initialization and learning coefficients. Neither
		  GLVQ nor FLVQ depends upon a choice for the update
		  neighborhood or learning rate distribution---these are
		  taken care of automatically.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bezdek93a,
  author	= {James C. Bezdek and Nikhil R. Pal and Eric C. K. Tsao},
  title		= {Two Generalizations of {K}ohonen Clustering},
  booktitle	= {Proc. of the Third Int. Workshop on Neural Networks and
		  Fuzzy Logic, Houston, Texas, NASA Conference Publication
		  10111},
  year		= {1993},
  editor	= {Christopher J. Culbert},
  volume	= {II},
  pages		= {199--226},
  publisher	= {NASA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bezdek93b,
  author	= {James C. Bezdek and Nikhil R. Pal},
  title		= {Prototype Generating Clustering Algorithms},
  booktitle	= {Proc. 5th IFSA World Congress '93---Seoul, Fifth Int.
		  Fuzzy Systems Association World Congress},
  year		= {1993},
  volume	= {I},
  pages		= {36--43},
  publisher	= {Korea Fuzzy Mathematics and Systems Society},
  address	= {Seoul, Korea},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bezdek93c,
  author	= {James C. Bezdek and Nikhil R. Pal},
  title		= {An Index of Topological Preservation and its Application
		  to Self-Organizing Feature Maps},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {III},
  pages		= {2435--2440},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {We discuss topological preservation under feature
		  extraction transformations. Transformations that preserve
		  the order of all distances in any neighborhood of vectors
		  in p-space are defined as metric topology preserving (MTP)
		  transformations. We give a necessary and sufficient
		  condition for this property in terms of Spearman's rank
		  correlation coefficient. A modification of Kohonen's
		  self-organizing feature map algorithm that extracts vectors
		  in q-space from data in p-space is given. Three methods are
		  empirically compared: principal components analysis:
		  Sammon's algorithm; and our extension of the
		  self-organizing feature map algorithm. Our MTP index shows
		  that the first two methods preserve distance ranks on six
		  data sets much more effectively than extended SOFM.},
  dbinsdate	= {oldtimer}
}

@Article{	  bezdek93d,
  author	= {Bezdek, J. and Pal, N. R. },
  title		= {Fuzzification of the \mbox{self-organizing} feature map:
		  will it work?},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1993},
  volume	= {2061},
  pages		= {142--62},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@Article{	  bezdek95a,
  author	= {James C. Bezdek and Nikhil R. Pal},
  title		= {A Note on Self-Organizing Semantic Maps},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1995},
  volume	= {6},
  number	= {5},
  pages		= {1029--1036},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {This paper discusses Kohonen's self-organizing semantic
		  map (SOSM). We show that augmentation and normalization of
		  numerical feature data as recommended for the SOSM is
		  entirely unnecessary to obtain semantic maps that exhibit
		  semantic similarities between objects represented by the
		  data. Visual displays of a small data set of 13 animals
		  based on principal components, Sammon's algorithm, and
		  Kohonen's (unsupervised) self-organizing feature map (SOFM)
		  possess exactly the same qualitative information as the
		  much more complicated SOSM display does.},
  dbinsdate	= {oldtimer}
}

@Article{	  bezdek95b,
  author	= {James C. Bezdek and Nikhil R. Pal},
  title		= {Two Soft Relatives of Learning Vector Quantization},
  journal	= {Neural Networks},
  year		= {1995},
  volume	= {8},
  number	= {5},
  pages		= {729--743},
  publisher	= {Elsevier Science Ltd},
  dbinsdate	= {oldtimer}
}

@Article{	  bezdek95c,
  author	= {Bezdek, J. C. and Pal, N. R. },
  title		= {Index of topological preservation for feature extraction},
  journal	= {Pattern Recognition},
  year		= {1995},
  volume	= {28},
  number	= {3},
  pages		= {381--91},
  month		= {March},
  abstract	= {This paper is about the ability of principal components
		  analysis, the Sammon algorithm, and an extension of the
		  Kohonen self-organizing feature map to preserve spatial
		  order during feature extraction on unlabeled data.
		  Transformations to q-space that preserve the order of all
		  pairwise distances in any set of vectors in p-space are
		  defined as metric topology preserving (MTP)
		  transformations. We give a necessary and sufficient
		  condition for this new property in terms of the Spearman
		  rank correlation coefficient. Unlike many other measures of
		  extracted feature quality, the MTP index is independent of
		  the extraction method. A modification of the Kohonen
		  self-organizing feature map algorithm that extracts vectors
		  in q-space from data in p-space is developed. The extent to
		  which principal components, Sammon's algorithm and our
		  extension of the self-organizing feature map (SOFM)
		  preserve the MTP property is discussed. Our MTP index shows
		  that the first two methods preserve distance ranks on seven
		  data sets much more effectively than extended SOFM.},
  dbinsdate	= {oldtimer}
}

@Article{	  bezdek95d,
  author	= {J. C. Bezdek and N. R. Pal and R. J. Hathaway and N. B.
		  Karayiannis},
  title		= {Some new competitive learning schemes},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1995},
  volume	= {2492},
  number	= {pt. 1},
  pages		= {538--49},
  note		= {(Applications and Science of Artificial Neural Networks
		  Conference Date: 17--21 April 1995 Conference Loc: Orlando,
		  FL, USA Conference Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bezdek96a,
  author	= {J. C. Bezdek and T. R. Reichherzer and G. Lim and Y.
		  Attikiouzel},
  title		= {Classification with multiple prototypes},
  booktitle	= {Proceedings of the Fifth IEEE International Conference on
		  Fuzzy Systems. FUZZ-IEEE '96},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {1},
  address	= {New York, NY, USA},
  pages		= {626--32},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bezdek99a,
  author	= {J. C. Bezdek and L. I. Kuncheva},
  title		= {Point prototype generation and classifier design},
  booktitle	= {Kohonen Maps},
  pages		= {71--96},
  publisher	= {Elsevier},
  year		= {1999},
  editor	= {Oja, E. and Kaski, S.},
  address	= {Amsterdam},
  annote	= {Nearest prototypes, self-organising feature maps,
		  classifier design,, pre-supervision, post-supervision,},
  dbinsdate	= {oldtimer}
}

@Article{	  bezerianos00a,
  author	= {Bezerianos, A. and Vladutu, L. and Papadimitriou, S.},
  title		= {Hierarchical state space partitioning with a network
		  self-organizing map for the recognition of {ST}-T segment
		  changes},
  journal	= {Medical and Biological Engineering and Computing},
  year		= {2000},
  volume	= {38},
  number	= {4},
  month		= {Jul},
  pages		= {406--415},
  organization	= {Univ of Patras},
  publisher	= {Peter Peregrinus Ltd},
  address	= {Stevenage},
  abstract	= {The problem of maximizing the performance of ST-T segment
		  automatic recognition for ischaemia detection is a
		  difficult pattern classification problem. The paper
		  proposes the network self-organizing map (NetSOM) model as
		  an enhancement to the Kohonen self-organized map (SOM)
		  model This model is capable of effectively decomposing
		  complex large-scale pattern classification problems into a
		  number of partitions, each of which is more manageable with
		  a local classification device. The NetSOM attempts to
		  generalize the regularization and ordering potential of the
		  basic SOM from the space of vectors to the space of
		  approximating functions. It becomes a device for the
		  ordering of local experts (i.e. independent neural
		  networks) over its lattice of neurons and for their
		  selection and co-ordination. Each local expert is an
		  independent neural network that is trained and activated
		  under the control of the NetSOM. This method is evaluated
		  with examples from the European ST-T database. The first
		  results obtained after the application of NetSOM to ST-T
		  segment change recognition show a significant improvement
		  in the performance compared with that obtained with
		  monolithic approaches, i.e. with single network types. The
		  basic SOM model has attained an average ischaemic beat
		  sensitivity of 73.6% and an average ischaemic beat
		  predictivity of 68.3%. The work reports and discusses the
		  improvements that have been obtained from the
		  implementation of a NetSOM classification system with both
		  multilayer perceptrons and radial basis function (RBF)
		  networks as local experts for the ST-T segment change
		  problem. Specifically, the NetSOM with multilayer
		  perceptrons (radial basis functions) as local experts has
		  improved the results over the basic SOM to an average
		  ischaemic beat sensitivity of 75.9% (77.7%) and an average
		  ischaemic beat predictivity of 72.5% (74.1%).},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bhandarkar01a,
  author	= {Bhandarkar, S. M. and Nammalwar, P.},
  title		= {Segmentation of multispectral {MR} images using a
		  hierarchical self-organizing map},
  booktitle	= {Proceedings of the IEEE Symposium on Computer-Based
		  Medical Systems},
  year		= {2001},
  editor	= {},
  volume	= {},
  pages		= {294--299},
  organization	= {Department of Computer Science, University of Georgia},
  publisher	= {},
  address	= {},
  abstract	= {The application of a hierarchical self-organizing map
		  (HSOM) to the problem of segmentation of multispectral
		  magneti resonance (MR) images is investigated. The HSOM is
		  composed of several layers of the self-organizing map (SOM)
		  organized in a pyramidal fashion. The SOM has been used for
		  the segmentation of multispectral MR images but the results
		  often suffer from undersegmentation and oversegmentation.
		  By combining the concepts of self-organization and
		  topographic mapping with multiscale image segmentation, the
		  HSOM is seen to overcome the major drawbacks of the SOM.
		  The segmentation results of the HSOM are compared with
		  those of the SOM and the k-means clustering algorithm on
		  multispectral MR images of the human brain representing
		  both, normal conditions and pathological conditions such as
		  multiple sclerosis. The multiscale segmentation results of
		  the HSOM are shown to have interesting consequences from
		  the viewpoint of clinical diagnosis of pathological
		  conditions.},
  dbinsdate	= {2002/1}
}

@Article{	  bhandarkar96a,
  author	= {S. M. Bhandarkar and J. Koh and Minsoo Suk},
  title		= {A hierarchical neural network and its application to image
		  segmentation},
  journal	= {Mathematics and Computers in Simulation},
  year		= {1996},
  volume	= {41},
  number	= {3--4},
  pages		= {337--55},
  note		= {(IMACS Symposium on Signal Processing Robotics and Neural
		  Networks Conference Date: April 1994 Conference Loc: Lille,
		  France)},
  abstract	= {The problem of image segmentation can be formulated as one
		  of vector quantization. Although self-organizing networks
		  with competitive learning are useful for vector
		  quantization, they, in their original single-layer
		  structure, are inadequate for image segmentation. This
		  paper proposes and describes a hierarchical self-organizing
		  neural network for image segmentation. The hierarchical
		  self-organizing feature map (HSOFM) which is an extension
		  of the traditional (single-layer) self-organizing feature
		  map (SOFM) is seen to alleviate the shortcomings of the
		  latter in the context of image segmentation. The problem of
		  image segmentation is formulated as one of vector
		  quantization and mapped onto the HSOFM. The HSOFM combines
		  the ideas of self-organization and topographic mapping with
		  those of multi-scale image segmentation. Experimental
		  results using intensity and range images bring out the
		  advantages of the HSOFM over the conventional SOFM.},
  dbinsdate	= {oldtimer}
}

@Article{	  bhandarkar97a,
  author	= {Bhandarkar, Suchendra M and Koh, Jean and Suk, Minsoo},
  title		= {Multiscale image segmentation using a hierarchical
		  \mbox{self-organizing} map},
  journal	= {Neurocomputing},
  year		= {1997},
  number	= {3},
  volume	= {14},
  pages		= {241--272},
  abstract	= {Multiscale structures and algorithms that unify the
		  treatment of local and global scene information are of
		  particular importance in image segmentation. Vector
		  quantization, owing to its versatility, has proved to be an
		  effective means of image segmentation. Although vector
		  quantization can be achieved using self-organizing maps
		  with competitive learning, self-organizing maps in their
		  original single-layer structure, are inadequate for image
		  segmentation. A hierarchical self-organizing neural network
		  for image segmentation is presented. The Hierarchical
		  Self-Organizing Map (HSOM) is an extension of the
		  conventional (single-layer) Self-Organizing Map (SOM). The
		  problem of image segmentation is formulated as one of
		  vector quantization and mapped onto the HSOM. By combining
		  the concepts of self-organization and topographic mapping
		  with those of multiscale image segmentation the HSOM
		  alleviates the shortcomings of the conventional {SOM} in
		  the context of image segmentation.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bharitkar01a,
  author	= {Bharitkar, S. and Filev, D.},
  title		= {An online learning vector quantization algorithm},
  booktitle	= {Proceedings of the Sixth International Symposium on Signal
		  Processing and its Applications. IEEE, Piscataway, NJ,
		  USA},
  year		= {2001},
  volume	= {2},
  pages		= {394--7},
  abstract	= {We propose an online learning algorithm for the learning
		  vector quantization (LVQ) approach in nonlinear supervised
		  classification. The advantage of this approach is the
		  ability of the LVQ to adjust its codebook vectors as new
		  patterns become available, so as to accurately model the
		  class representation of the patterns. Moreover this
		  algorithm does not significantly increase the computational
		  complexity over the original LVQ algorithm.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bi94a,
  author	= {Hao Bi and Guangguo Bi and Yimin Mao},
  title		= {Stochastically Competitive Learning Algorithm for Vector
		  Quantizer Design},
  pages		= {622--626},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {vector quantization, modification, comparison},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bi94b,
  author	= {Hao Bi and Guangguo Bi and Yimin Mao},
  title		= {Globally Optimal Vector Quantizer Design Using
		  Stochastically Competitive Learning Algorithm},
  booktitle	= {Proc. Int. Symp. on Speech, Image Processing and Neural
		  Networks},
  year		= {1994},
  volume	= {II},
  pages		= {650--653},
  organization	= {{IEEE} Hong Kong Chapter of Signal Processing},
  address	= {Hong Kong},
  annote	= {application, vector quantization, optimization,
		  modification},
  dbinsdate	= {oldtimer}
}

@Article{	  bian99a,
  author	= {Bian, J.~C. and Chen, H.~B. and Yang, P.~C. and Lu,
		  D.~R.},
  title		= {Retrievals of over-Ocean Precipitable Water from {S}sm/{I}
		  by {SOM} Network Model},
  journal	= {Chinese Science Bulletin},
  year		= {1999},
  volume	= {44},
  number	= {11},
  pages		= {1038--1041},
  dbinsdate	= {oldtimer}
}

@Article{	  biebelmann96a,
  author	= {E. Biebelmann and M. Koppen and B. Nickolay},
  title		= {Practical applications of neural networks in texture
		  analysis},
  journal	= {Neurocomputing},
  year		= {1996},
  volume	= {13},
  number	= {2--4},
  pages		= {261--79},
  note		= {(3rd International Conference on Fuzzy Logic, Neural Nets
		  and Soft Computing (IIZUKA'94) Conf. Date: 1--7 Aug. 1994
		  Conf. Loc: Iizuka, Japan Conf. Sponsor: Int. Fuzzy Syst.
		  Assoc. ; Int. Neural Network Soc. ; Japan Soc. Fuzzy Theory
		  \& Syst. ; et al)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bieler94a,
  author	= {Bieler, K. and Glavitsch, H. },
  title		= {Evaluation of different AI-methods for fault diagnosis in
		  power systems},
  booktitle	= {ISAP '94. International Conference on Intelligent System
		  Application to Power Systems},
  year		= {1994},
  editor	= {Hertz, A. and Holen, A. T. and Rault, J. -C. },
  volume	= {1},
  pages		= {209--16},
  organization	= {Eidgenossische Tech. Hochschule, Zurich, Switzerland},
  publisher	= {EC2},
  address	= {Nanterre Cedex, France},
  dbinsdate	= {oldtimer}
}

@Article{	  bienfait94a,
  author	= {Bienfait, B. },
  title		= {Applications of high-resolution \mbox{self-organizing}
		  maps to retrosynthetic and {QSAR} analysis},
  journal	= {Journal of Chemical Information and Computer Sciences},
  year		= {1994},
  volume	= {34},
  number	= {4},
  pages		= {890--8},
  month		= {July-Aug},
  dbinsdate	= {oldtimer}
}

@Article{	  bienfait97a,
  author	= {B. Bienfait and J. Gasteiger},
  title		= {Checking the projection display of multivariate data with
		  colored graphs},
  journal	= {Journal of Molecular Graphics \& Modelling},
  year		= 1997,
  volume	= 15,
  pages		= {203--215,254--258},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bigus94a,
  author	= {Joseph P. Bigus},
  title		= {Applying Neural Networks to Computer System Performance
		  Tuning},
  pages		= {2442--2447},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, initialization, hybrid},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bing91a,
  author	= {Z. Bing and E. Grant},
  title		= {A neural network approach to adaptive state-space
		  partitioning},
  booktitle	= {Proc. IEEE Int. Symp. on Intelligent Control},
  year		= {1991},
  pages		= {180--183},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  binks91a,
  author	= {David L. Binks and Nigel M. Allinson},
  title		= {Financial Data Recognition and Prediction using Neural
		  Networks},
  booktitle	= {Artificial Neural Networks},
  year		= {1991},
  editor	= {T. Kohonen and K. M{\"{a}}kisara and O. Simula and J.
		  Kangas},
  volume	= {II},
  pages		= {1709--1712},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  month		= {},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bishop91a,
  author	= {J. M. Bishop and R. J. Mitchell},
  title		= {Neural networks---an introduction},
  booktitle	= {Proc. IEE Colloquium on 'Neural Networks for Systems:
		  Principles and Applications' (Digest No. 019)},
  year		= {1991},
  pages		= {1--3},
  organization	= {IEE},
  publisher	= {IEE},
  address	= {London, UK},
  x		= {In this paper, neural networks are introduced, . . .
		  Kohonen networks; . . . },
  dbinsdate	= {oldtimer}
}

@TechReport{	  bishop96a,
  author	= {Christopher M. Bishop and Markus Svens{\'e}n and
		  Christopher K. I. Williams},
  title		= {{GTM}: A Principled Alternative to the Self-Organizing
		  Map},
  institution	= {Neural Computing Research Group, Aston University},
  year		= 1996,
  number	= {NCRG/96/015},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bishop96b,
  author	= {C. M. Bishop and M. Svensen and C. K. I. Williams},
  title		= {{GTM}: a principled alternative to the
		  \mbox{self-organizing} map},
  booktitle	= {Artificial Neural Networks---ICANN 96. 1996 International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1996},
  editor	= {C. {von der Malsburg} and W. {von Seelen} and J. C.
		  Vorbruggen and B. Sendhoff},
  address	= {Berlin, Germany},
  pages		= {165--70},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bishop97a,
  author	= {Christopher M. Bishop and Markus Svensen and Christopher
		  K. I. Williams},
  title		= {Magnification factors for the {SOM} and {GTM} algorithms},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {333--338},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bishop97b,
  author	= {Christopher M. Bishop and Markus Svens{\'e}n and
		  Christopher K. I. Williams},
  title		= {{GTM}: A Principled Alternative to the Self-Organizing
		  Map},
  booktitle	= {Advances in Neural Information Processing Systems 9},
  publisher	= {The MIT Press},
  year		= 1997,
  editor	= {Michael C. Mozer and Michael I. Jordan and Thomas
		  Petsche},
  pages		= {354--360},
  address	= {Cambridge, MA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bishop98a,
  author	= {C. M. Bishop},
  title		= {Latent variables, topographic mappings and data
		  visualization},
  booktitle	= {Neural Nets WIRN-VIETRI-97. Proceedings of the 9th Italian
		  Workshop on Neural Nets},
  publisher	= {Springer-Verlag London},
  year		= {1998},
  editor	= {M. Marinaro and R. Tagliaferri},
  address	= {London, UK},
  pages		= {3--32},
  dbinsdate	= {oldtimer}
}

@Article{	  bishop98b,
  author	= {Bishop, Christopher M and Svensen, Markus and Williams,
		  Christopher KI},
  title		= {Developments of the generative topographic mapping},
  journal	= {Neurocomputing},
  year		= {1998},
  number	= {1},
  volume	= {21},
  pages		= {203--224},
  abstract	= {The generative topographic mapping (GTM) model was
		  introduced by Bishop et al. as a probabilistic
		  re-formulation of the self-organizing map (SOM). It offers
		  a number of advantages compared with the standard {SOM},
		  and has already been used in a variety of applications. In
		  this paper we report on several extensions of the GTM,
		  including an incremental version of the EM algorithm for
		  estimating the model parameters, the use of local subspace
		  models, extensions to mixed discrete and continuous data,
		  semi-linear models which permit the use of high-dimensional
		  manifolds whilst avoiding computational intractability,
		  Bayesian inference applied to hyper-parameters, and an
		  alternative framework for the GTM based on Gaussian
		  processes. All of these developments directly exploit the
		  probabilistic structure of the GTM, thereby allowing the
		  underlying modelling assumptions to be made explicit. They
		  also highlight the advantages of adopting a consistent
		  probabilistic framework for the formulation of pattern
		  recognition algorithms.},
  dbinsdate	= {oldtimer}
}

@Article{	  bishop98c,
  author	= {Christopher M. Bishop and Markus Svens{\'e}n and
		  Christopher K. I. Williams},
  title		= {{GTM}: The Generative Topographic Mapping},
  journal	= {Neural Computation},
  year		= 1998,
  volume	= 10,
  pages		= {215--234},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  black93a,
  author	= {Black, J. V. },
  title		= {Comparison of the performance of vector quantiser training
		  algorithms},
  booktitle	= {Third International Conference on Artificial Neural
		  Networks},
  year		= {1993},
  pages		= {71--5},
  organization	= {Defence Res. Agency, Malvern, UK},
  publisher	= {IEE},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@TechReport{	  blackmore92a,
  author	= {Justine Blackmore and Risto Miikkulainen},
  title		= {Incremental grid growing: encoding high-dimensional
		  structure into a two-dimensional feature map},
  institution	= {University of Texas at Austin},
  number	= {TR AI92--192},
  address	= {Austin, TX},
  year		= {1992 },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blackmore93a,
  author	= {Justine Blackmore and Risto Miikkulainen},
  title		= {Incremental Grid Growing: Encoding High-Dimensional
		  Structure into a Two-Dimensional Feature Map},
  booktitle	= {Proc. ICNN'93, International Conference on Neural
		  Networks},
  year		= {1993},
  volume	= {I},
  pages		= {450--455},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InCollection{	  blackmore95a,
  author	= {J. Blackmore and R. Miikkulainen},
  title		= {Visualizing high-dimensional structure with the
		  incremental grid growing neural network},
  booktitle	= {Machine Learning. Proceedings of the Twelfth International
		  Conference on Machine Learning},
  publisher	= {Morgan Kaufmann Publishers},
  year		= {1995},
  editor	= {A. Prieditis and S. Russell},
  address	= {San Francisco, CA, USA},
  pages		= {55--63},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blanch00a,
  author	= {Blanch Perez del Notario C. and Malanda Trigueros, A.},
  title		= {Kohonen-{GLA} algorithms for efficient vector
		  quantization},
  booktitle	= {ICECS 2000. 7th IEEE International Conference on
		  Electronics, Circuits and Systems. IEEE, Piscataway, NJ,
		  USA},
  year		= {2000},
  volume	= {2},
  pages		= {671--4},
  abstract	= {Two new methods for vector quantization (VQ) codebook
		  design are presented, consisting of a sequential
		  combination of two well-known VQ design techniques: the
		  Kohonen self organizing feature maps and the generalized
		  Lloyd algorithm. The combined methods profit from the best
		  features of their constituents achieving a superior
		  performance to either of them.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  blanchet92a,
  author	= {Max Blanchet and Shuji Yoshizawa and Nobuyuki Okudaira and
		  {Shun-ichi} Amari},
  title		= {Self-Adaptive System for Automatic Sleep Cycle Recognition
		  Using Heart Rate. Application for a Biological Rythme
		  Dependent Alarm Clock},
  booktitle	= {Proc. 7'th Symp. on Biological and Physiological
		  Engineering},
  year		= {1992},
  pages		= {171--174},
  publisher	= {Toyohashi University of Technology},
  address	= {Toyohashi, Japan},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blanchet93a,
  author	= {Max Blanchet and Shuji Yoshizawa and {Shun-ichi} Amari},
  title		= {Modified {K}ohonen's Self-Organizing Feature Map and Its
		  Application to Automatic Sleep Cycle Recognition},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {III},
  pages		= {2476--2479},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {In order to increase the adaptive ability of the Kohonen's
		  self-organizing feature map algorithm, we modify the
		  updating rule of its partition boundary. The modified
		  algorithm, together with the Genetic Algorithm, is applied
		  to the detection of the REM (Rapid Eye Movement) cycle of
		  the sleep, which is said to be related to comfortableness
		  of the sleep. We used the heart-rate alone (not relying on
		  complicated EEG analysis) and succeeded to achieve a high
		  detection accuracy.},
  dbinsdate	= {oldtimer}
}

@Article{	  blanco00a,
  author	= {Blanco, A. and Delgado, M. and Pegalajar, M. C.},
  title		= {Extracting rules from a (fuzzy/crisp) recurrent neural
		  network using a self-organizing map},
  journal	= {International Journal of Intelligent Systems},
  year		= {2000},
  volume	= {15},
  number	= {7},
  month		= {},
  pages		= {595--621},
  organization	= {Universidad de Granada},
  publisher	= {John Wiley \& Sons Inc},
  address	= {New York, NY},
  abstract	= {Although the extraction of symbolic knowledge from trained
		  feedforward neural networks has been widely studied,
		  research in recurrent neural networks (RNN) has been more
		  neglected, even though it performs better in areas such as
		  control, speech recognition, time series prediction, etc.
		  Nowadays, a subject of particular interest is (crisp/fuzzy)
		  grammatical inference, in which the application of these
		  neural networks has proven to be suitable. In this paper,
		  we present a method using a self-organizing map (SOM) for
		  extracting knowledge from a recurrent neural network able
		  to infer a (crisp/fuzzy) regular language. Identification
		  of this language is done only from a (crisp/fuzzy) example
		  set of the language.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  blanco99a,
  author	= {Blanco, I. D. and {Diez Gonzalez}, A. B. and {Cuadrado
		  Vega}, A. A. and {Enguita Gonzalez}, J.},
  title		= {{RBF} approach for trajectory interpolation in
		  \mbox{self-organizing} map based condition monitoring},
  booktitle	= {1999 7th IEEE International Conference on Emerging
		  Technologies and Factory Automation. Proceedings ETFA '99.
		  },
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {2},
  pages		= {1003--10},
  abstract	= {Self-organizing map trajectory interpolation using radial
		  basis functions is proposed for process condition
		  monitoring. Self-organizing maps are a powerful means to
		  visualize the state of a process. However, a quantization
		  process takes place when input vectors are mapped onto the
		  grid (output) space. The discrete nature of the output
		  space gives rise to a considerable loss of information
		  which is particularly harmful to detect incipient faults
		  which are often revealed by drifts or slight changes in the
		  state trajectory. A continuous projection from the input
		  space onto the {SOM} grid space using RBF-based
		  interpolation is proposed and tested with experimental
		  data.},
  dbinsdate	= {oldtimer}
}

@Article{	  blasco00a,
  author	= {Blasco, J. A. and Fueyo, N. and Dopazo, C. and Chen, J.
		  Y.},
  title		= {A self-organizing-map approach to chemistry representation
		  in combustion applications},
  journal	= {COMBUSTION THEORY AND MODELLING},
  year		= {2000},
  volume	= {4},
  number	= {1},
  month		= {MAR},
  pages		= {61--76},
  abstract	= {Several alternative techniques have been proposed in the
		  literature in order to avoid the CPU-intensive numerical
		  integration of the thermochemical equations in the
		  simulation of combustion processes. The present paper
		  introduces a new approach, which is based on two artificial
		  neural-network (ANN) paradigms, namely the self-organizing
		  map (SOM) and the multilayer perceptron (MLP). The SOM is
		  first employed for the automatic partitioning of the
		  thermochemical space into subdomains. Then, a specialized
		  MLP is trained in order to fit the thermochemical points
		  belonging to a given subdomain. The presented strategy is
		  tested on a partially stirred reactor (PaSR) with a reduced
		  methane-air mechanism, and encouraging results are
		  reported. The relatively modest CPU-time and memory
		  requirements of the method make the SOM-MLP approach a
		  promising technique for the inclusion of large chemical
		  mechanisms in the context of complex applications, such as
		  the multidimensional simulation of combustion.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  blayo90a,
  author	= {F. Blayo and C. Lehmann},
  title		= {A systolic implementation of the self organization
		  algorithm},
  booktitle	= {Proc. INNC'90, Int. Neural Network Conf. },
  year		= {1990},
  organization	= {Thomsom; SUN; British Computer Society ; et al},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blayo91a,
  author	= {F. Blayo and P. Demartines},
  title		= {Data analysis: how to compare {K}ohonen neural networks to
		  other techniques?},
  booktitle	= {Proc. IWANN'91, Int. Workshop on Artificial Neural
		  Networks},
  year		= {1991},
  editor	= {A. Prieto},
  pages		= {469--476},
  publisher	= {Springer},
  address	= {Berlin, Heidelberg},
  dbinsdate	= {oldtimer}
}

@Article{	  blayo92a,
  author	= {F. Blayo and P. Demartines},
  title		= {{K}ohonen algorithms. {A}pplication to the analysis of
		  economic data},
  journal	= {Bull. des Schweizerischen Elektrotechnischen Vereins {\&}
		  des Verbandes Schweizerischer Elektrizit{\"{a}}tswerke},
  year		= {1992},
  volume	= {83},
  number	= {5},
  pages		= {23--26},
  note		= {(in French)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blight93a,
  author	= {Blight, D. C. and McLeod, R. D. },
  title		= {Self-organizing {K}ohonen maps for {FPGA} placement},
  booktitle	= {Field-Programmable Gate Arrays: Architectures and Tools
		  for Rapid Prototyping. Second International Workshop on
		  Field Programmable Logic and Applications},
  year		= {1993},
  editor	= {Grunbacher, H. and Hartenstein, R. W. },
  pages		= {88--95},
  organization	= {Dept. of Electr. \& Comput. Eng. , Manitoba Univ. ,
		  Winnipeg, Man. , Canada},
  publisher	= {Springer-Verlag},
  address	= {Berlin, Germany},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blonda93a,
  author	= {P. Blonda and G. Pasquariello and J. Smith},
  title		= {Comparison of Backpropagation, Cascade-Correlation and
		  {K}ohonen Algorithms for Cloud Retrieval},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {II},
  pages		= {1231--1234},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {The expected high volume imagery data from NASA Mission to
		  the Planet Earth (e.g. one terabyte for EOS AM) is one of
		  the target application areas for automated cloud retrieval,
		  and more generally for automated image classification. We
		  used the backpropagation (BP), the Cascade-Correlation (CC)
		  and Kohonen self-organizing map (SOM) neural network
		  architectures for cloud retrieval from satellite imagery.
		  We have used a simple scene (a mixed scene containing only
		  cloud and ocean). This simple scene allows us to evaluate
		  the accuracy of the classification (and the trend in
		  misclassifications). better than a complicated scene. BP
		  and CC performed at the same accuracy level. CC was
		  slightly more efficient than BP in terms of the number
		  epochs. BP requires the user to set the number of hidden
		  units. CC demonstrated a built -in flexibility in terms of
		  the (variable) number of hidden units necessary to
		  accomplish the learning phase of the algorithm. The {SOM}
		  algorithm, was slightly less accurate, do to its
		  unsupervised nature, than CC and BP for our test data ( 97%
		  accuracy versus 99% accuracy level for BP and CC). This
		  study shows that for simple scenes, which are abundant in
		  global monitoring satellite imagery, a simple
		  pixel-by-pixel or 3-by-3 window approaches provide high
		  accuracy classification without using complicated
		  contextual information.},
  dbinsdate	= {oldtimer}
}

@Article{	  blonda93b,
  author	= {Blonda, P. and Carella, A. and DeBlasi, R. and Dicuonzo,
		  F. and LaForgia, V. and Milella, D. Pasquariello, G. and
		  Satalino, G.},
  title		= {Neural network modular system for object classification in
		  brain {MR} images},
  journal	= {Artificial Intelligence in Medicine},
  year		= {1993},
  number	= {},
  volume	= {10},
  pages		= {477--480},
  abstract	= {In this paper we present a modular system for segmentation
		  and classification of brain Magnetic Resonance (MR) images.
		  It consists of two Neural Network architectures, a Self
		  Organizing Map and a Multilayer Perceptron. The objective
		  of the work is to evaluate the effectiveness of a Neural
		  Network approach for automatic recognition of anatomic
		  structures in {MR} images. The system has been useful to
		  discriminate 20 different anatomic classes.},
  dbinsdate	= {oldtimer}
}

@Article{	  blonda94a,
  author	= {Blonda, P. and La Forgia, V. and Pasquariello, G. and
		  Satalino, G. },
  title		= {Feature extraction and pattern classification for remotely
		  sensed data analysis by a modular neural system},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1994},
  volume	= {2315},
  pages		= {48--55},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blonda95a,
  author	= {Blonda, P. and Bennardo, A. and la Forgia, V. and
		  Satalino, G. },
  title		= {Modular neural system, based on a fuzzy clustering
		  network, for classification},
  booktitle	= {1995 International Geoscience and Remote Sensing
		  Symposium, IGARSS '95. Quantitative Remote Sensing for
		  Science and Applications},
  year		= {1995},
  editor	= {Stein, T. I. },
  volume	= {1},
  pages		= {449--51},
  organization	= {IESI-CNR, Bari, Italy},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  blonda96a,
  author	= {P. Blonda and A. Bennardo and G. Satalino and V. la
		  Forgia},
  title		= {Application of the unsupervised Fuzzy {K}ohonen Clustering
		  Network for remote sensed data segmentation},
  booktitle	= {Proceedings of the WILF '95. Italian Workshop on Fuzzy
		  Logic 1995. New Trends in Fuzzy Logic},
  publisher	= {World Scientific},
  year		= {1996},
  editor	= {A. Bonarini and D. Mancini and F. Masulli and A.
		  Petrosino},
  address	= {Singapore},
  pages		= {143--50},
  dbinsdate	= {oldtimer}
}

@Article{	  blonda96b,
  author	= {P. Blonda and V. la Forgia and G. Pasquariello and G.
		  Satalino},
  title		= {Feature extraction and pattern classification of remote
		  sensing data by a modular neural system},
  journal	= {Optical Engineering},
  year		= {1996},
  volume	= {35},
  number	= {2},
  pages		= {536--42},
  dbinsdate	= {oldtimer}
}

@Article{	  blonda96c,
  author	= {P. Blonda and A. Bennardo and G. Satalino and G.
		  Pasquariello and R. {De Blasi} and D. Milella},
  title		= {Fuzzy neural network based segmentation of multispectral
		  magnetic resonance brain images},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1996},
  volume	= {2761},
  pages		= {146--53},
  note		= {(Applications of Fuzzy Logic Technology III Conf. Date:
		  10--12 April 1996 Conf. Loc: Orlando, FL, USA Conf.
		  Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@Article{	  blonda96d,
  author	= {P. Blonda and A. Bennardo and G. Pasquariello and G.
		  Satalino and V. la Forgia},
  title		= {Application of the fuzzy {K}ohonen clustering network to
		  remote sensed data processing},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1996},
  volume	= {2761},
  pages		= {119--29},
  note		= {(Applications of Fuzzy Logic Technology III Conf. Date:
		  10--12 April 1996 Conf. Loc: Orlando, FL, USA Conf.
		  Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  blonda97a,
  author	= {P. Blonda and A. Bennardo and G. Satalino},
  title		= {Neuro-fuzzy processing of remote sensed data},
  booktitle	= {Neural Nets WIRN VIETRI-96. Proceedings of the 8th Italian
		  Workshop on Neural Nets},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {M. Marinaro and R. Tagliaferri},
  address	= {London, UK},
  pages		= {153--63},
  dbinsdate	= {oldtimer}
}

@Article{	  blonda97b,
  author	= {P. Blonda and A. Baraldi and G. Bafunno and G. Satalino
		  and G. Ria},
  title		= {Experimental comparison of {FOSART} and {F {LVQ} } in a
		  remotely sensed image classification task},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1997},
  volume	= {3165},
  pages		= {113--22},
  note		= {(Applications of Soft Computing Conf. Date: 28--29 July
		  1997 Conf. Loc: San Diego, CA, USA Conf. Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  blonda99a,
  author	= {Blonda, P. and Satalino, G. and Wasowski, J. and Parise,
		  M. and Baraldi A and Refice A},
  title		= {Neural techniques for {SAR} intensity and coherence data
		  classification},
  booktitle	= {Proceedings-of-the-SPIE
		  --The-International-Society-for-Optical-Engineering.
		  vol.3871},
  year		= {1999},
  volume	= {3871},
  pages		= {374--81},
  abstract	= {Previously it has been proved that combined analysis of
		  SAR intensity and interferometric correlation images is a
		  valuable tool in classification tasks where traditional
		  techniques such as crisp thresholding schemes and classical
		  maximum likelihood classifiers have been employed. In this
		  work, developed in the framework of the ESA A03--320
		  project titled Application of ERS data to landslide
		  activity monitoring in the southern Apennines, Italy, our
		  goal is to investigate: 1) usefulness of SAR
		  interferometric correlation information in mapping areas
		  with diffuse erosional activity, including landslides; and
		  2) effectiveness of soft computing techniques in the
		  combined analysis of SAR intensity and interferometric
		  correlation images. Two neural classifiers are selected
		  from the literature. The first classifier is a one-stage
		  error-driven multilayer perceptron (MLP) and the second
		  classifier is a two-stage hybrid (TSH) learning system,
		  consisting of a sequence of an unsupervised data-driven
		  first stage with a supervised error-driven second stage.
		  The TSH unsupervised first stage is implemented as either:
		  a) the on-line learning, dynamic-sizing, dynamic-linking
		  fully self organizing simplified adaptive resonance theory
		  (FOSART) clustering model; b) the batch-learning,
		  static-sizing, no-linking fuzzy learning vector
		  quantization (FLVQ) algorithm; or c) the on-line learning,
		  static-sizing, static-linking self-organizing map (SOM).
		  The input data set consists of three SAR ERS-1/ERS-2 tandem
		  pair images depicting an area featuring slope instability
		  phenomena in the Campanian Apennines of Southern Italy.
		  From each tandem pair, four pixel-based features are
		  extracted: the backscattering mean intensity, the
		  interferometric coherence, the backscattering intensity
		  texture and the backscattering intensity change. Our
		  classification task is focused on the discrimination of
		  land cover types useful for hazard evaluation, i.e.,
		  evaluation of areas affected by erosion.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  blumel94a,
  author	= {Blumel, R. },
  title		= {Application of {K}ohonen's \mbox{self-organizing}
		  artificial neural networks to {PWM} inverter drives},
  booktitle	= {IECON '94. 20th International Conference on Industrial
		  Electronics, Control and Instrumentation},
  year		= {1994},
  volume	= {2},
  pages		= {1242--6},
  organization	= {Univ. der Bundeswehr Munchen, Neubiberg, Germany},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  bobrovskii96a,
  author	= {A. L. Bobrovskii and V. V. Efirnov},
  title		= {Intelligent information systems with parallel data
		  processing},
  journal	= {Elektronnoe Modelirovanie},
  year		= {1996},
  volume	= {18},
  number	= {1},
  pages		= {24--8},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bock96a,
  author	= {Hans H. Bock},
  title		= {Simultaneous Visualization and Classification Methods as
		  an Alternative to {K}ohonen's Neural Networks},
  booktitle	= {Classification and Multivariate Graphics: Models, Software
		  and Applications},
  year		= 1996,
  editor	= {Hans-Joachim Mucha and Hans-Hermann Bock},
  number	= {Report No. 10},
  series	= {Weierstrass-Institut f{\"u}r Angewandte Analysis und
		  Stochastik},
  address	= {Berlin},
  pages		= {15--23},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  boda94a,
  author	= {P{\'{e}}ter Boda and Gy{\"{o}}rgy G. Vass},
  title		= {Neural Networks and Fuzzy Systems in Speech Processing:
		  Applications to Voiced/Unvoiced Decision},
  editor	= {Christer Carlsson and Timo J{\"{a}}rvi and Tapio Reponen},
  number	= {12},
  series	= {Conf. Proc. of Finnish Artificial Intelligence Society},
  pages		= {47--54},
  booktitle	= {Proc. Conf. on Artificial Intelligence Res. in Finland},
  year		= {1994},
  publisher	= {Finnish Artificial Intelligence Society},
  address	= {Helsinki, Finland},
  annote	= {application, pattern recognition, comparison},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  boda95a,
  author	= {Boda, P. P. },
  title		= {Robust voiced/unvoiced speech classification with
		  \mbox{self-organizing} maps},
  booktitle	= {1995 IEEE Symposium on Circuits and Systems},
  year		= {1995},
  volume	= {2},
  pages		= {1516--19},
  organization	= {Acoust. Lab. , Helsinki Univ. of Technol. , Espoo,
		  Finland},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  boddy94a,
  author	= {L. Boddy and A. M. Gimblett and C. W. Morris and J. E. M.
		  Mordue},
  title		= {Neural network analysis of fungal spore morphometric data
		  for identification of species in the genus Pestalotiopsis},
  booktitle	= {Intelligent Engineering Systems Through Artificial Neural
		  Networks.},
  publisher	= {ASME},
  volume	= {4},
  year		= {1994},
  editor	= {C. H. Dagli and B. R. Fernandez and J. Ghosh and R. T. S.
		  Kumara},
  address	= {New York, NY, USA},
  pages		= {605--12},
  dbinsdate	= {oldtimer}
}

@Article{	  bode01a,
  author	= {Bode, M. and Freyd, O. and Fischer, J. and
		  Niedernostheide, F.-J. and Schulze, H.-J.},
  title		= {Hybrid hardware for a highly parallel search in the
		  context of learning classifiers},
  journal	= {Artificial Intelligence},
  year		= {2001},
  volume	= {130},
  number	= {1},
  month		= {July 2001},
  pages		= {75--84},
  organization	= {Westfalische Wilhelms-Univ. Munster, Institut fur
		  Angewandte Physik},
  publisher	= {},
  address	= {},
  abstract	= {Based on a comparison of input data with a set of
		  prototypes, classifier systems identify the most
		  appropriate representative for a given sample pattern. One
		  remarkable classifier is Kohonen's Self-Organizing Map and
		  the related learning vector quantizer, as these algorithms
		  are highly parallel. For real-time applications the
		  classifier search may be one of the time critical
		  processes. We discuss specialized hardware being able to
		  execute such a search in a fully parallel manner. Also the
		  learning and updating of prototypes is performed in
		  parallel controlled by a propagating front. Finally, we
		  present experimental results concerning an unsupervised
		  learning vector quantizer (LVQ) and a self-organizing map
		  (SOM) obtained from our thyristor-based analog-digital
		  hybrid system. },
  dbinsdate	= {2002/1}
}

@InProceedings{	  bodruzzaman95a,
  author	= {M. Bodruzzaman and S. Zein-Sabatto and O. Omitowoju and M.
		  Malkani},
  title		= {Electromyographic {(EMG)} Signal Decomposition using
		  {K}ohonen Neural Net and Wavelet Network},
  volume	= {II},
  pages		= {854--862},
  booktitle	= {Proc. WCNN'95, World Congress on Neural Networks},
  year		= {1995},
  organization	= {INNS},
  dbinsdate	= {oldtimer}
}

@Article{	  boehm94a,
  author	= {Boehm, K. and Broll, W. and Sokolewicz, M. },
  title		= {Dynamic gesture recognition using neural networks; a
		  fundament for advanced interaction construction},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1994},
  volume	= {2177},
  pages		= {336--46},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  boehme94a,
  author	= {H. -J. Boehme and U. -D. Braumann and H. -M. Gross},
  title		= {A Neural Network Architecture for Sensory Controlled
		  Internal Simulation},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {II},
  pages		= {1189--1192},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, knowledge database, internal state},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bogdan01a,
  author	= {Bogdan, M. and Rosenstiel, W.},
  title		= {Detection of cluster in self-organizing maps for
		  controlling a prostheses using nerve signals},
  booktitle	= {9th European Symposium on Artificial Neural Networks.
		  ESANN'2001. Proceedings. D-Facto, Evere, Belgium},
  year		= {2001},
  volume	= {},
  pages		= {131--6},
  abstract	= {In order to control a prostheses by means of biological
		  nerve signals, a self-organizing map (SOM) was used to
		  classify nerve signals recorded by a regeneration type
		  neurosensor. The trained SOM contains information about the
		  relation between the recorded nerve signal and the winning
		  neuron of the SOM. Classes of nerve signals fired by
		  defined axons can be found in a cluster on the SOM. In
		  order to control a prostheses, the clusters on the SOM must
		  be assigned to an action of the prostheses. Since medical
		  staff are usually not experienced at identifying the
		  clusters within an SOM, we have developed Clusot, an
		  algorithm that automatically defines clusters within SOMs.
		  After a short introduction to the project, we present the
		  signal processing of the project. We focus on the automatic
		  detection of clusters within a trained SOM using Clusot.
		  Clusot is explained within the context of the project in
		  question.},
  dbinsdate	= {2002/1}
}

@Article{	  bogdan92a,
  author	= {Bogdan, A. and Meadows, H. E. },
  title		= {{K}ohonen neural network for image coding based on
		  iteration transformation theory},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1992},
  volume	= {1766},
  pages		= {425--36},
  annote	= {A conference paper in journal},
  abstract	= {Iterated transformation theory (ITT), also known as
		  fractal coding, is a relatively new block compression
		  method which removes redundancies between different scale
		  representations of the uncompressed signal. In ITT coding
		  we are looking for a piecewise continuous mapping from the
		  space of all images with the same support onto itself which
		  has a close approximation of the desired image as a unique
		  fixed point. The mapping is then the code for the image,
		  and for decoding we iterate the mapping on any initial
		  image, orders of magnitude faster than encoding. We have
		  reduced the computational load of finding the piecewise
		  continuous transformation by using a self-organizing
		  feature map (SOFM) artificial neural network which finds
		  similar features in different resolution representations of
		  the image. The patterns are mapped onto a two-dimensional
		  array of formal neurons forming a code book similar to
		  vector quantization (VQ) coding. We use the (SOFM) ordering
		  properties by searching for mapping not only to the best
		  feature match neuron but also to its neighbors in the
		  network. In this paper we describe the ITT-SOFM algorithm
		  and its software implementation with application to image
		  coding of still gray images. Computer simulations show
		  compression results comparable to or better than
		  state-of-the-art VQ coders, and computational complexity
		  better than most of the well known clustering algorithms.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  boggess94a,
  author	= {Boggess, J. E. , III and Nation, P. B. and Harmon, M. E.
		  },
  title		= {Compression of color information in digitized images using
		  an artificial neural network},
  booktitle	= {Proceedings of the IEEE 1994 National Aerospace and
		  Electronics Conference NAECON 1994},
  year		= {1994},
  volume	= {2},
  pages		= {772--8},
  organization	= {Dept. of Comput. Sci. , Mississippi State Univ. , MS,
		  USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  boggess94b,
  author	= {Boggess, J. E. , III and Nation, P. B. and Harmon, M. E.
		  },
  title		= {Using artificial neural networks for data compression of
		  color information in digitized images},
  booktitle	= {Proceedings of the 32nd Annual Southeast Conference},
  year		= {1994},
  editor	= {Cordes, D. W. and Vrbsky, V. },
  pages		= {298--304},
  organization	= {Dept. of Comput. Sci. , Mississippi State Univ. , MS,
		  USA},
  publisher	= {ACM},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bohn98a,
  author	= {ChristianA. Bohn},
  title		= {{K}ohonen Feature Mapping through Graphics Hardware},
  booktitle	= {Proc. JCIS'98},
  publisher	= {Association for Intelligent Machinery, Inc},
  year		= 1998,
  editor	= {Paul P. Wang},
  volume	= {II},
  pages		= {64--67},
  abstract	= {This work describes the utilization of the inherent
		  parallelism of commonly available hardware graphics
		  accelerators for the realization of the Kohonen feature
		  map. The result is an essential reduction of computing time
		  compared to standard software implementations.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bojer01a,
  author	= {Bojer, T. and Hammer, B. and Schunk, D. and Tluk von
		  Toschanowitz K.},
  title		= {Relevance determination in learning vector quantization},
  booktitle	= {9th European Symposium on Artificial Neural Networks.
		  ESANN'2001. Proceedings. D-Facto, Evere, Belgium},
  year		= {2001},
  volume	= {},
  pages		= {271--6},
  abstract	= {We propose a method to automatically determine the
		  relevance of the input dimensions of a learning vector
		  quantization (LVQ) architecture during training. The method
		  is based on Hebbian learning and introduces weighting
		  factors of the input dimensions which are automatically
		  adapted to the specific problem. The benefits are twofold.
		  On the one hand, the incorporation of relevance factors in
		  the LVQ architecture increases the overall performance of
		  the classification and adapts the metric to the specific
		  data used for training. On the other hand, the method
		  induces a pruning algorithm, i.e. an automatic detection of
		  the input dimensions which do not contribute to the overall
		  classifier. Hence we obtain a possibly more efficient
		  classification and we gain insight to the role of the data
		  dimensions.},
  dbinsdate	= {2002/1}
}

@InCollection{	  bologna94a,
  author	= {G. Bologna and C. Pellegrini},
  title		= {Internal knowledge analysis in a feed-forward neural
		  network},
  booktitle	= {Neural Networks in Biomedicine. Proceedings of the
		  Advanced School of the Italian Biomedical Physics
		  Association},
  publisher	= {World Scientific},
  year		= {1994},
  editor	= {F. Masulli and P. G. Morasso and A. Schenone},
  address	= {Singapore},
  pages		= {37--56},
  dbinsdate	= {oldtimer}
}

@Article{	  bonabeau02a,
  author	= {Bonabeau, Eric},
  title		= {Graph multidimensional scaling with self-organizing maps},
  journal	= {Information Sciences},
  year		= {2002},
  volume	= {143},
  number	= {1--4},
  month		= {June },
  pages		= {159--180},
  organization	= {Icosystem Corporation},
  publisher	= {Elsevier Science Inc.},
  address	= {},
  abstract	= {Self-organizing maps (SOM) are unsupervised, competitive
		  neural networks used to project high-dimensional data onto
		  a low-dimensional space. In this paper it is shown that SOM
		  can be used to perform multidimensional scaling (MDS) on
		  graphs. The SOM-based approach is applied to two families
		  of random graphs and three real-world networks. },
  dbinsdate	= {2002/1}
}

@Article{	  bonabeau98a,
  author	= {E. Bonabeau and F. Henaux},
  title		= {\mbox{Self-organizing} maps for drawing large graphs},
  journal	= {Information Processing Letters},
  year		= {1998},
  volume	= {67},
  number	= {4},
  pages		= {177--84},
  dbinsdate	= {oldtimer}
}

@Article{	  bonabeau98b,
  author	= {E. Bonabeau and F. Henaux},
  title		= {Monte {C}arlo Partitioning of Graphs with a Self
		  Organizing Map},
  journal	= {International Journal of Modern Physics C},
  volume	= {9},
  pages		= {1107--1119},
  year		= {1998},
  dbinsdate	= {oldtimer}
}

@Article{	  bonnet95a,
  author	= {N. Bonnet},
  title		= {Preliminary Investication of two Methods for the Automatic
		  Handling of Multivariate Maps in Microanalysis},
  journal	= {Ultramicroscopy},
  year		= {1995},
  volume	= {57},
  number	= {1},
  pages		= {17--27},
  publisher	= {Elsevier},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bors94a,
  author	= {Adrian G. Bor{\c{s}} and I. Pitas},
  title		= {Robust Estimation for Radial Basis Functions},
  booktitle	= {Proc. NNSP'94, IEEE Workshop on Neural Networks for Signal
		  Processing},
  year		= {1994},
  month		= {September},
  organization	= {IEEE},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  pages		= {105--114},
  annote	= {application, modification},
  dbinsdate	= {oldtimer}
}

@Article{	  bors96a,
  author	= {A. G. Bors and I. Pitas},
  title		= {Median radial basis function neural network},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1996},
  volume	= {7},
  number	= {6},
  pages		= {1351--64},
  dbinsdate	= {oldtimer}
}

@Article{	  bortolan02a,
  author	= {Bortolan, G. and Pedrycz, W.},
  title		= {An interactive framework for an analysis of {ECG}
		  signals},
  journal	= {ARTIFICIAL INTELLIGENCE IN MEDICINE},
  year		= {2002},
  volume	= {24},
  number	= {2},
  month		= {FEB},
  pages		= {109--132},
  abstract	= {In this study, we introduce and discuss a development of a
		  highly interactive and user-friendly environment for an ECG
		  signal analysis. The underlying neural architecture being a
		  crux of this environment comes in the form of a
		  self-organizing map. This map helps discover a structure in
		  a set of ECG patterns and visualize a topology of the data.
		  The role of the designer is to choose from some already
		  visualized regions of the self-organizing map characterized
		  by a significant level of data homogeneity and substantial
		  difference from other regions. In the sequel, the regions
		  are described by means of information granules-fuzzy sets
		  that are essential in the characterization of the main
		  relationships existing in the ECG data. The study
		  introduces an original method of constructing membership
		  functions that incorporates class membership as an
		  important factor affecting changes in membership grades.
		  The study includes a comprehensive descriptive modeling of
		  highly dimensional ECG data.},
  dbinsdate	= {2002/1}
}

@Book{		  bose96a,
  author	= {Bose, S. and Murthy, C.A},
  title		= {New {LVQ} Model. Technical rept.},
  year		= {1996},
  abstract	= {A new LVQ model has been proposed here. An exponential
		  membership function has been considered in this regard. The
		  performance of the new model in relation to other existing
		  models has been studied experimentally with the help of an
		  artificial data set as well as IRIS data. Finally the
		  proposed algorithm is applied on a satellite image data.
		  The proposed model has been found to provide satisfactory
		  results with all these data sets.},
  dbinsdate	= {oldtimer}
}

@Article{	  bottazzi90a,
  author	= {C. Bottazzi},
  title		= {Neuro-computers},
  journal	= {Informazione Elettronica},
  year		= {1990},
  volume	= {18},
  number	= {10},
  pages		= {21--27},
  month		= {October},
  note		= {(in Italian)},
  x		= {. . . author discusses progress in the technology of the
		  application of neutral networks and . . . with special
		  reference to work being done by T. Kohonen . . . },
  dbinsdate	= {oldtimer}
}

@InCollection{	  bouchired98a,
  author	= {S. Bouchired and M. Ibnkahla and D. Roviras and F.
		  Castanie},
  title		= {Equalization of satellite mobile communication channels
		  using combined \mbox{self-organizing} maps and {RBF}
		  networks},
  booktitle	= {Proceedings of the 1998 IEEE International Conference on
		  Acoustics, Speech and Signal Processing, ICASSP '98},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {6},
  address	= {New York, NY, USA},
  pages		= {3377--9},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bouton91a,
  author	= {C. Bouton and M. Cottrell and J. -C. Fort and G.
		  Pag{\`{e}}s},
  title		= {Self-organization and convergence of the {K}ohonen
		  algorithm},
  booktitle	= {Probabilit{\'{e}}s Num{\'{e}}riques},
  publisher	= {INRIA},
  address	= {Paris, France},
  year		= {1991},
  editor	= {N. Bouleau and D. Talay},
  chapter	= {V. 2},
  pages		= {163--180},
  dbinsdate	= {oldtimer}
}

@TechReport{	  bouton92a,
  author	= {C. Bouton and G. Pag{\`{e}}s},
  title		= {Self-organization and convergence of the
		  \mbox{\mbox{one-dimensional}} {K}ohonen algorithm with non
		  uniformly distributed stimuli (version 2)},
  institution	= {Laboratoire de Probabilit{\'{e}}s, Universit{\'{e}} Paris
		  VI},
  year		= {1992},
  address	= {Paris, France},
  month		= {April},
  dbinsdate	= {oldtimer}
}

@TechReport{	  bouton92b,
  author	= {C. Bouton and G. Pag{\`{e}}s},
  title		= {Convergence in distribution of the
		  \mbox{\mbox{one-dimensional}} {K}ohonen algorithms when the
		  stimuli are not uniform},
  institution	= {Laboratoire de Probabilit{\'{e}}s, Universit{\'{e}} Paris
		  VI},
  year		= {1992},
  address	= {France},
  month		= {April},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bouton92c,
  author	= {Catherine Bouton and Gilles Pag{\`e}s},
  title		= {Auto-Organisation de l'Algorithme de {K}ohonen en
		  Dimension~1},
  year		= {1992},
  booktitle	= {Proc. Workshop `Aspects Theoriques des Reseaux de
		  Neurones'},
  editor	= {M. Cottrell and M. Chaleyat-Maurel},
  publisher	= {Universit\'e Paris~{I}},
  address	= {Paris, France},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bouton92d,
  author	= {Catherine Bouton and Gilles Pag{\`e}s},
  title		= {Convergence p. s. et en Loi de l'Algorithme de {K}ohonen
		  en Dimension~1},
  year		= {1992},
  booktitle	= {Proc. of the workshop `Aspects Theoriques des Reseaux de
		  Neurones'},
  editor	= {M. Cottrell and M. Chaleyat-Maurel},
  publisher	= {Universit\'e Paris~{I}},
  address	= {Paris, France},
  dbinsdate	= {oldtimer}
}

@Article{	  bouton93a,
  author	= {Catherine Bouton and Gilles Pag{\`{e}}s},
  title		= {Self-Organization and a. s. convergence of the
		  \mbox{\mbox{one-dimensional}} {K}ohonen algorithm with
		  non-uniformly distributed stimuli},
  journal	= {Stochastic Processes and Their Applications},
  year		= {1993},
  volume	= {47},
  pages		= {249--274},
  dbinsdate	= {oldtimer}
}

@Article{	  bouton94a,
  author	= {C. Bouton and G. Pag\`es},
  title		= {Convergence in distribution of the
		  \mbox{\mbox{one-dimensional}} {{K}ohonen} algorithms when
		  the stimuli are not uniform},
  journal	= {Advances in Applied Probability},
  volume	= {26},
  pages		= {80--103},
  year		= {1994},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bovbel00a,
  author	= {Bovbel, E. E. and Kheidorov, I. E. and Kotlyar, M. E.},
  title		= {Speaker identification using autoregressive hidden Markov
		  models and adaptive vector quantisation},
  booktitle	= {Text, Speech and Dialogue. Third International Workshop,
		  TSD 2000. Proceedings (Lecture Notes in Artificial
		  Intelligence Vol.1902). Springer-Verlag, Berlin, Germany},
  year		= {2000},
  volume	= {},
  pages		= {207--10},
  abstract	= {Wide-frequency spectral analysis, autoregressive hidden
		  Markov models (ARHMM) and self-organising neural networks
		  (SOM) have been used for high accuracy speaker features
		  modelling. The initial ARHMM parameters estimation based on
		  Kalman filter is proposed. The five-keyword speaker
		  identification system has been built and tested. The
		  experiments show that this approach provides high accuracy
		  of speaker identification even if the same words are
		  pronounced by different speakers.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bowles99a,
  author	= {Bowles, Jeffrey and Clamons, Dean and Gillis, David and
		  Palmadesso, Peter and Antoniades, John and Baumback, Mark
		  and Daniel, Mark and Grossmann, John and Haas, Daniel and
		  Skibo, Jeffrey},
  title		= {New results from the ORASIS/NEMO compression algorithm},
  booktitle	= {Proceedings of SPIE---The International Society for
		  Optical Engineering},
  year		= {1999},
  volume	= {3753},
  pages		= {226--234},
  abstract	= {We present results from an improved ORASIS (Optical
		  Real-time Adaptive Spectral Identification System)
		  hyperspectral-data compression-algorithm that is being
		  implemented on the Naval EarthMap Observer (NEMO)
		  satellite. The algorithm is shown to produce results that
		  are statistically improved from previous findings. To
		  augment the statistical testing, the re-inflated data are
		  run through analysis programs such as unsupervised
		  classification. ORASIS compression is a series of
		  algorithms. The first algorithm, the exemplar selector
		  process (ESP), is a variation of Learned Vector
		  Quantization (LVQ) that builds up a relatively small set of
		  spectra to represent the full data set. Subsequent
		  algorithms find approximate endmembers for the exemplar set
		  and project the set into the space defined by the
		  endmembers. Both the ESP and the projection process
		  contribute to the compression of the data. The obtainable
		  compression ratios vary with scene content and other
		  factors but ratios between 10:1 and 30:1 are possible. The
		  compressed data format is designed to allow direct access
		  to individual pieces of the data without reinflation of the
		  entire data set. Details of the hardware implementation of
		  the Imagery On-Board Processor (IOBP) of NEMO is discussed,
		  as well as the use of the compressed data on the ground.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  boznar97a,
  author	= {M. Boznar},
  title		= {Pattern selection strategies for a neural network-based
		  short term air pollution prediction model},
  booktitle	= {Proceedings. Intelligent Information Systems. IIS'97},
  publisher	= {IEEE Computer Society},
  year		= {1997},
  editor	= {H. Adeli},
  address	= {Los Alamitos, CA, USA},
  pages		= {340--4},
  dbinsdate	= {oldtimer}
}

@InCollection{	  braccini97a,
  author	= {G. Braccini and L. Edenbrandt and M. Lagerholm and C.
		  Peterson and O. Rauer and R. Rittner and L. Sornmo},
  title		= {\mbox{Self-organizing} maps and {H}ermite functions for
		  classification of {ECG} complexes},
  booktitle	= {Computers in Cardiology 1997},
  publisher	= {IEEE},
  year		= {1997},
  address	= {New York, NY, USA},
  pages		= {425--8},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bradburn89a,
  author	= {David S. Bradburn},
  title		= {Reducing Transmission Error Effects Using a
		  Self-Organizing Network},
  booktitle	= {Proc. IJCNN'89, International Joint Conference on Neural
		  Networks},
  year		= {1989},
  volume	= {II},
  pages		= {531--537},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  bradburn91a,
  author	= {Bradburn, D. S.},
  title		= {Self-organization of non-numeric data sets.},
  journal	= {International Joint Conference on Neural Networks},
  year		= {1991},
  number	= {},
  volume	= {},
  pages		= {37--42},
  abstract	= {A topology-preserving map can be constructed by simulated
		  annealing. This map shares some useful properties with
		  those generated by classical self-organization, but it does
		  not require that the input data have numeric values. An
		  application to mail routing is presented, showing that
		  self-organization can be accomplished given only a local
		  adjacency list. A string-matching application shows the
		  extension of the algorithm to cases where a scalar distance
		  between input points is available. The map in the latter
		  application, like a Kohonen map, exhibits local smoothness
		  permitting shortcuts to the nearest-neighbor search.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  branston99a,
  author	= {Branston, N. M. and {El Deredy}, W.},
  title		= {Capturing movement responses of single cells in the human
		  basal ganglia using hidden {M}arkov models},
  booktitle	= {ICANN99. Ninth International Conference on Artificial
		  Neural Networks (IEE Conf. Publ. No.470)},
  year		= {1999},
  publisher	= {IEE},
  address	= {London, UK},
  volume	= {1},
  pages		= {395--400},
  abstract	= {We used hidden {M}arkov models (HMMs) to characterise the
		  firing activity of single neurons in the basal ganglia
		  recorded during voluntary limb movements made by patients
		  undergoing surgery for Parkinson's disease. The parameters
		  of these HMMs were then used as input patterns for training
		  the generative topographic mapping (GTM) algorithm. Data
		  were strongly clustered in the GTM latent space, and
		  separated both by brain region and by movement type, when
		  as few as two states were modelled in the HMM. Our results
		  support the hypotheses that (1) cells in these brain
		  regions are not involved in the preparation or execution of
		  a single type of movement but rather participate to some
		  extent in many different movements, and (2) neural networks
		  in these regions operate as systems with distinct and
		  relatively stable configurations of activity.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  brauer89a,
  author	= {P. Brauer and P. Knagenhjelm},
  title		= {Infrastructure in {K}ohonen maps},
  booktitle	= {Proc. ICASSP-89, International Conference on Acoustics,
		  Speech and Signal Processing},
  year		= {1989},
  pages		= {647--650},
  address	= {},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  brauer91a,
  author	= {P. Brauer and P. Hedelin and D. Huber and P. Knagenhjelm},
  title		= {Probability based optimization for network classifiers},
  booktitle	= {Proc. ICASSP-91, International Conference on Acoustics,
		  Speech and Signal Processing},
  year		= {1991},
  volume	= {I},
  pages		= {133--136},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  x		= {. . . The authors propose a probability based optimization
		  (PBO) algorithm . . . The performance of this algorithm has
		  been systematically evaluated using a MAP (maximum a
		  posteriori) estimator and a Kohonen (1988) map. },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  brause90a,
  author	= {R. Brause},
  title		= {Optimal performance and storage requirements of
		  neighbourhood-conserving mappings for robot control},
  booktitle	= {Proc. INNC'90, Int. Neural Network Conference},
  year		= {1990},
  volume	= {I},
  pages		= {221--224},
  organization	= {Thomsom; SUN; British Computer Society ; et al},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  brause90b,
  author	= {R. Brause},
  title		= {Optimal information distribution and performance in
		  neighbourhood-conserving maps for robot control},
  booktitle	= {Proc. 2nd Int. IEEE Conference on Tools for Artificial
		  Intelligence},
  year		= {1990},
  pages		= {451--456},
  organization	= {IEEE},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA},
  dbinsdate	= {oldtimer}
}

@Article{	  brause92a,
  author	= {R. Brause},
  title		= {Optimal Information Distribution and Performance in
		  Neighbourhood-Conserving Maps for Robot Control},
  journal	= {Int. J. Computers and Artificial Intelligence},
  year		= {1992},
  volume	= {11},
  number	= {2},
  pages		= {173--199},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  brause94a,
  author	= {R{\"{u}}diger W. Brause},
  title		= {An Approximation Network with Maximal Transinformation},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {I},
  pages		= {701--704},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, approximation, modification},
  dbinsdate	= {oldtimer}
}

@Article{	  brause96a,
  author	= {Brause, R. W. },
  title		= {Sensor encoding using lateral inhibited self-organized
		  cellular neural networks},
  journal	= {Neural Networks},
  year		= {1996},
  volume	= {9},
  number	= {1},
  pages		= {99--120},
  abstract	= {The paper focuses on the division of the sensor field into
		  subsets of sensor events and proposes the linear
		  transformation with the smallest achievable error for
		  reproduction: the transform coding approach using the
		  principal component analysis (PCA). For the implementation
		  of the PCA, this paper introduces a new symmetrical,
		  lateral inhibited neural network model, proposes an
		  objective function for it and deduces the corresponding
		  learning rules. The necessary conditions for the learning
		  rate and the inhibition parameter for balancing the
		  crosscorrelations vs the autocorrelations are computed. The
		  simulation reveals that an increasing inhibition can speed
		  up the convergence process in the beginning slightly. In
		  the remaining part of the paper, the application of the
		  network in picture encoding is discussed. Here, the use of
		  non-completely connected networks for the self-organized
		  formation of templates in cellular neural networks is
		  shown. It turns out that the self-organizing Kohonen map is
		  just the non-linear, first order approximation of a general
		  self-organizing scheme. Hereby, the classical transform
		  picture coding is changed to a parallel, local model of
		  linear transformation by locally changing sets of
		  self-organized eigenvector projections with overlapping
		  input receptive fields. This approach favours an effective,
		  cheap implementation of sensor encoding directly on the
		  sensor chip.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  breazu99a,
  author	= {Breazu, M. and Toderean, G. and Volovici, D. and Iridon,
		  M.},
  title		= {Speeding up fractal image compression by working in
		  {K}arhunen-{L}oeve transform space},
  booktitle	= {IJCNN'99. International Joint Conference on Neural
		  Networks. Proceedings},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {4},
  pages		= {2694--7},
  abstract	= {The main weakness of fractal image compression is its long
		  encoding time needed to search the entire domain pool to
		  find the best domain-range mapping. To solve the problem,
		  some solutions were proposed but most of them do not employ
		  neural networks (only the use of Kohonen {SOM} for
		  clustering was reported). The paper proposes a new method
		  based on {K}arhunen-{L}oeve transform (PCA networks), which
		  attempts to use neural networks' well-known adaptability in
		  order to find a good feature vector for a block.
		  Performance regarding network generality, quantization of
		  the transform coefficients, comparison with DCT and kd-tree
		  search, was explored. Results prove that the proposed
		  method slightly outperforms state-of-the-art methods.},
  dbinsdate	= {oldtimer}
}

@Book{		  breidegrad00a,
  author	= {Björn Breidegard},
  title		= {En datorexekverbar modell för lärande},
  publisher	= {Licentiatuppsats Certec, LTH},
  year		= {2000},
  number	= {1},
  dbinsdate	= {2002/1}
}

@Article{	  breining00a,
  author	= {Breining, Christina},
  title		= {State detection for hands-free telephone sets by means of
		  fuzzy {LVQ} and {SOM}},
  journal	= {Signal Processing},
  year		= {2000},
  volume	= {80},
  number	= {7},
  month		= {Jul},
  pages		= {1361--1372},
  organization	= {Technische Universitaet Darmstadt},
  publisher	= {Elsevier Science Publishers B.V.},
  address	= {Amsterdam},
  abstract	= {The step gain of adaptive algorithms for acoustic echo
		  cancellation should vary with time in dependence of the
		  far-end signal, the local signal, and the adaptation
		  quality. It can be described as state-dependent, which is
		  also the case for the control parameters of other
		  algorithms in a hands-free telephone set, such as adaptive
		  loss control or noise reduction. Most of the known state
		  detection methods can only distinguish between subsets of
		  the states, and are unreliable for certain state
		  transitions, so that they should be combined for improved
		  reliability of the state detection. In this contribution,
		  state detection is performed by a fuzzy learning vector
		  quantization/self-organizing map approach to combine
		  several state detectors. The step gain is inferred from
		  state information using fuzzy logic concepts. Finally,
		  adaptation results are compared to the best conventional
		  step-gain control method used in the complete structure.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  breton95a,
  author	= {S. Breton and J. P. Urban and H. Kihl},
  title		= {A Recursive Sensorimotor Map-based Algorithm for the
		  Learning of Saccades},
  volume	= {II},
  pages		= {406--409},
  booktitle	= {Proc. WCNN'95, World Congress on Neural Networks},
  year		= {1995},
  organization	= {INNS},
  dbinsdate	= {oldtimer}
}

@InCollection{	  briscoe97a,
  author	= {G. Briscoe and T. Caelli},
  title		= {Learning temporal sequences in recurrent
		  \mbox{self-organising} neural nets},
  booktitle	= {Advanced Topics in Artificial Intelligence. 10th
		  Australian Joint Conference on Artificial Intelligence,
		  AI'97. Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {A. Sattar},
  address	= {Berlin, Germany},
  pages		= {427--35},
  dbinsdate	= {oldtimer}
}

@InCollection{	  brockmann97a,
  author	= {D. Brockmann and H. -U. Bauer and M. Riesenhuber and T.
		  Geisel},
  title		= {{SOM}-model for the development of oriented receptive
		  fields and orientation maps from non-oriented ON-center
		  OFF-center inputs},
  booktitle	= {Artificial Neural Networks---ICANN '97. 7th International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {W. Gerstner and A. Germond and M. Hasler and J. -D.
		  Nicoud},
  address	= {Berlin, Germany},
  pages		= {207--12},
  dbinsdate	= {oldtimer}
}

@Article{	  brosse01a,
  author	= {Brosse, S. and Giraudel, J. L. and Lek, S.},
  title		= {Utilisation of non-supervised neural networks and
		  principal component analysis to study fish assemblages},
  journal	= {ECOLOGICAL MODELLING},
  year		= {2001},
  volume	= {146},
  number	= {1--3},
  month		= {DEC 1},
  pages		= {159--166},
  abstract	= {Kohonen self-organizing maps (SOM) belong to the
		  non-supervised artificial neural network modelling methods.
		  It typically displays a high dimensional data set in a
		  lower dimensional space. In this way, that method can be
		  considered as a non- linear surrogate to the principal
		  component analysis (PCA). In order to test the efficiency
		  of SOM on complex ecological data gathered in the natural
		  environment, we made a comparison between PCA and SOM
		  capabilities to analyse the spatial occupancy of several
		  European freshwater fish species in the littoral zone of a
		  large French lake. The same data matrix consisting of 710
		  samples and 15 species was analysed using PCA and SOM. Both
		  methods provided insights on the major trends in fish
		  spatial occupancy. However, a more detailed analysis showed
		  that only SOM was able to reliably visualise the entire
		  fish assemblage in a two dimensional space (i.e. both
		  dominant and scarce species). On the contrary PCA provided
		  irrelevant ecological information for some species. These
		  drawbacks were afforded to data heterogeneity, scarce
		  species being poorly represented on the PCA plane. These
		  results led us to conclude that SOM constitute a more
		  reliable data representation method than PCA when complex
		  ecological data sets are used. },
  dbinsdate	= {2002/1}
}

@InProceedings{	  brozek98a,
  author	= {Brozek, J. and Kulczycki, J. and Szpyra, W. and Tylek,
		  W.},
  title		= {Designing the m-loops electric power distribution networks
		  using artificial neural networks},
  booktitle	= {Engineering Benefits from Neural Networks. Proceedings of
		  the International Conference EANN '98},
  publisher	= {Systems Engineering Association, Turku, Finland},
  year		= {1998},
  volume	= {},
  pages		= {261--4},
  abstract	= {The use of artificial neural networks for designing m-loop
		  electric power distribution networks is presented. The task
		  is to connect transformer stations (with given locations
		  and loads) into m loops (m Hamilton cycles) in order to
		  minimize a given aim function, subject to technical
		  constraints. The problem of designing the described network
		  is similar to the travelling salesman problem and is a
		  NP-hard problem. The described task can be solved in three
		  stages: dividing a transformer station set into m subsets
		  using a competitive neural network with winner-takes-all
		  (WTA) learning rule; connecting each subset element using
		  the Kohonen self-organizing map (SOM), after connecting
		  every subset to the main feeding point (MFP); improving the
		  obtained result. As the optimization criterion the minimal
		  annual cost of the electric power network (i.e., fixed
		  cost+variable cost+cost of undelivered energy) was chosen.
		  The definition of the problem as well as a short algorithm
		  and some results of tests performed on m-loop electric
		  power network models are presented.},
  dbinsdate	= {oldtimer}
}

@InBook{	  bruckner02a,
  author	= {Bernd Br{\"u}ckner and Thomas Wesarg},
  editor	= {Udo Seiffert and Lakhmi C. Jain},
  title		= {Self-Organizing Neural Networks---Recent Advances and
		  Applications},
  chapter	= {Modeling Speech Processing and Recognition in the Auditory
		  System Using the Multilevel Hypermap Architecture},
  publisher	= {Physica-Verlag Heidelberg},
  year		= {2002},
  key		= {},
  volume	= {78},
  number	= {},
  series	= {Studies in Fuzziness and Soft Computing},
  type		= {},
  address	= {},
  edition	= {},
  month		= {},
  pages		= {145--64},
  note		= {},
  annote	= {},
  dbinsdate	= {2002/1}
}

@InProceedings{	  bruckner92a,
  author	= {B. Br{\"{u}}ckner and M. Franz and A. Richter},
  title		= {A Modified Hypermap Architecture for Classification of
		  Biological Signals},
  booktitle	= {Artificial Neural Networks, 2},
  year		= {1992},
  editor	= {I. Aleksander and J. Taylor},
  volume	= {II},
  pages		= {1167--1170},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  month		= {},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bruckner93a,
  author	= {B. Br{\"{u}}ckner and W. Zander},
  title		= {Classification of Speech using a Modified {H}ypermap
		  Architecture},
  booktitle	= {Proc. WCNN'93, World Congress on Neural Networks},
  year		= {1993},
  volume	= {III},
  pages		= {75--78},
  organization	= {{INNS}},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bruckner93b,
  author	= {B. Br{\"{u}}ckner and W. Zander},
  title		= {Neurobiological Modelling and Structured Neural Networks},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {43--46},
  publisher	= {Springer},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bruckner95a,
  author	= {B. Br{\"{u}}ckenr and T. Wesarg and C. Blumenstein},
  title		= {Improvements of the modified Hypermap Architecture for
		  Speech Recognition},
  volume	= {V},
  pages		= {2891--2895},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bruckner97a,
  author	= {B. Br{\"u}ckner},
  title		= {Improvements in the analysis of structured data with the
		  Multilevel Hypermap Architecture},
  booktitle	= {Progress in Connectionsist-Based Information Systems.
		  Proceedings of the 1997 International Conference on Neural
		  Information Processing and Intelligent Information
		  Systems},
  publisher	= {Springer},
  year		= 1997,
  editor	= {Nikola Kasabov and Robert Kozma and Kitty Ko and Robert
		  O'Shea and George Coghill and Tom Gedeon},
  volume	= {1},
  address	= {Singapore},
  pages		= {342--345},
  dbinsdate	= {oldtimer}
}

@Article{	  bruske95a,
  author	= {Bruske, J. and Sommer, G. },
  title		= {Dynamic cell structure learns perfectly topology
		  preserving map},
  journal	= {Neural Computation},
  year		= {1995},
  volume	= {7},
  number	= {4},
  pages		= {845--65},
  month		= {July},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bruske95b,
  author	= {Bruske, J. and Sommer, G. },
  title		= {Dynamic cell structures},
  booktitle	= {Advances in Neural Information Processing Systems 7},
  year		= {1995},
  editor	= {Tesauro, G. and Touretzky, D. and Leen, T. },
  pages		= {497--504},
  organization	= {Dept. of Cognitive Syst. , Kiel Univ. , Germany},
  publisher	= {MIT Press},
  address	= {Cambridge, MA, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  bruske96a,
  author	= {J. Bruske and I. Ahrns and G. Sommer},
  title		= {Practicing Q-learning},
  booktitle	= {4th European Symposium on Artificial Neural Networks,
		  ESANN '96. Proceedings},
  publisher	= {D Facto},
  year		= {1996},
  editor	= {M. Verleysen},
  address	= {Brussels, Belgium},
  pages		= {25--30},
  dbinsdate	= {oldtimer}
}

@Article{	  bruske97a,
  author	= {J. Bruske and M. Hansen and L. Riehn and G. Sommer},
  title		= {Biologically inspired calibration-free adaptive saccade
		  control of a binocular camera-head},
  journal	= {Biological Cybernetics},
  year		= {1997},
  volume	= {77},
  number	= {6},
  pages		= {433--46},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bryant93a,
  author	= {Bryant, B. D. and Gowdy, J. N. },
  title		= {Speaker-independent voiced-stop-consonant recognition
		  using a block-windowed neural network architecture},
  booktitle	= {Proceedings SSST '93 The Twenty-Fifth Southeastern
		  Symposium on System Theory},
  year		= {1993},
  pages		= {400--4},
  organization	= {Dept. of Electr. \& Comput. Eng. , Clemson Univ. , SC,
		  USA},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  brzakovic92a,
  author	= {Brzakovic, D. and Wang, D. and Beck, H. },
  title		= {Modular neural network architecture for flaw
		  classification},
  booktitle	= {Southcon /92. Conference Record},
  year		= {1992},
  pages		= {315--19},
  organization	= {Dept. of Electr. \& Comput. Eng. , Tennessee Univ. ,
		  Knoxville, TN, USA},
  publisher	= {Electron. Conventions Manage},
  address	= {Los Angeles, CA, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  buckle01a,
  author	= {Buckle, M. and Strey, A.},
  title		= {Specification and simulation of neural networks using
		  epsilo{NN}},
  journal	= {Neural-Network-World},
  year		= {2001},
  volume	= {11},
  pages		= {73--89},
  abstract	= {The language EpsiloNN allows a high-level specification of
		  arbitrary neural network structures. It is especially
		  designed for the automatic generation of simulation code,
		  which can run efficiently on different parallel computer
		  architectures. In this paper some applications of EpsiloNN
		  are presented. First, the basic syntactical and semantical
		  aspects of the language are described briefly. Then the
		  EpsiloNN specifications of a popular multilayer perceptron
		  (MLP) and of a more complex hybrid LVQ/RBF neural network
		  architecture are presented. Further features of the
		  language are explained by example.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  budinich94a,
  author	= {Marco Budinich and John G. Taylor},
  title		= {On the Ordering Conditions for {S}elf-{O}rganizing
		  {M}aps},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {I},
  pages		= {347--349},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {analysis, ordering considerations, multidimensional},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  budinich94b,
  author	= {Marco Budinich},
  title		= {A {S}elf-{O}rganizing Neural Network for the Traveling
		  Salesman Problem that is Competitive with Simulated
		  Annealing},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {I},
  pages		= {358--361},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, TSP, optimization},
  dbinsdate	= {oldtimer}
}

@Article{	  budinich95a,
  author	= {Marco Budinich and John G. Taylor},
  journal	= {Neural Computation},
  title		= {On the Ordering Conditions for {S}elf-{O}rganizing
		  {M}aps},
  pages		= {284--289},
  volume	= {7},
  number	= {2},
  year		= {1995},
  dbinsdate	= {oldtimer}
}

@Article{	  budinich95b,
  author	= {Marco Budinich},
  journal	= {Neural Computation},
  title		= {Sorting with {S}elf-{O}rganizing {M}aps},
  pages		= {1188--1190},
  volume	= {7},
  number	= {6},
  year		= {1995},
  dbinsdate	= {oldtimer}
}

@Article{	  budinich96a,
  author	= {Marco Budinich},
  journal	= {Neural Computation},
  title		= {A {S}elf-{O}rganizing Neural Network for the Traveling
		  Salesman Problem That is Competitive with Simulated
		  Annealing},
  pages		= {416--424},
  volume	= {8},
  number	= {2},
  year		= {1996},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buessler00a,
  author	= {Buessler, J. L. and Urban, J. P.},
  title		= {Neurobiology suggests the design of modular architectures
		  for neural control},
  booktitle	= {IEEE International Conference on Intelligent Robots and
		  Systems},
  year		= {2000},
  editor	= {},
  volume	= {1},
  pages		= {64--69},
  organization	= {TROP Research Group, University of Mulhouse},
  publisher	= {},
  address	= {},
  abstract	= {The existence of modular structures in the biological
		  world strongly suggests that the training of this kind of
		  structures is actually feasible. It is a key indication for
		  the development of neural networks applications, especially
		  in the field of robotics. Indeed, a single network can only
		  efficiently treat problems with few independent variables;
		  the combining of several networks is necessary to address
		  more complex tasks. We investigate learning techniques and
		  show that using a particular form of architecture can ease
		  the training of a modular structure: a bi-directional
		  structure that allows combining several neural networks.
		  The approach is illustrated with Kohonen's self-organizing
		  maps for a robotic visual servoing task.},
  dbinsdate	= {2002/1}
}

@Article{	  buessler02a,
  author	= {Buessler, Jean-Luc and Urban, Jean-Philippe and Gresser,
		  Julien},
  title		= {Additive composition of supervised self-organizing maps},
  journal	= {Neural Processing Letters},
  year		= {2002},
  volume	= {15},
  number	= {1},
  month		= {February },
  pages		= {9--20},
  organization	= {TROP Research Group},
  publisher	= {},
  address	= {},
  abstract	= {The learning of complex relationships can be decomposed
		  into several neural networks. The modular organization is
		  determined by prior knowledge of the problem that permits
		  to split the processing into tasks of small dimensionality.
		  The sub-tasks can be implemented with neural networks,
		  although the learning examples cannot be used anymore to
		  supervise directly each of the networks. This article
		  addresses the problem of learning in a modular context,
		  developing in particular additive compositions. A simple
		  rule allows defining efficient training, and combining, for
		  example, several Supervised-SOM networks. This technique is
		  important because it introduces interesting generalizations
		  in many modular compositions, permitting data fusion or
		  sequential combinations of neural networks.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  buessler95a,
  author	= {J. L. Buessler and D. Kuhn and J. P. Urban},
  title		= {Learning Self-Organizing Maps Using Input-Output
		  Associations Applied to Robotics},
  volume	= {II},
  pages		= {384--388},
  booktitle	= {Proc. WCNN'95, World Congress on Neural Networks},
  year		= {1995},
  organization	= {INNS},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buessler99a,
  author	= {Buessler, J. L. and Kara, R. and Wira, P. and Kihl, H. and
		  Urban, J. P.},
  title		= {Multiple \mbox{self-organizing} maps to facilitate the
		  learning of visuo-motor correlations},
  booktitle	= {IEEE SMC'99 Conference Proceedings. 1999 IEEE
		  International Conference on Systems, Man, and Cybernetics.
		  },
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {3},
  pages		= {470--5},
  abstract	= {This paper presents an application of bi-directional
		  neural modularity: a chaining of several self-organizing
		  maps (SOM) is used to represent the motor and sensorial
		  position correlations of a robotic platform. Two active
		  cameras follow the movements of a robot manipulator in 3-D
		  space. The mapping of image positions and camera
		  orientations into arm angular joint positions can be
		  learned by a neural network. However, decomposing the
		  problem and using several neural networks turns out to be a
		  better way. In our approach, the neural modules do not need
		  to be adapted independently. Based on the principle of
		  bi-directionality, the modular architecture can be adapted
		  globally, using the sensor-motor data directly.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  buhmann89a,
  author	= {Joachim Buhmann and Robert Divko and Klaus Schulten},
  title		= {On Sparsely Coded Associative Memories},
  booktitle	= {Neural Networks from Models to Applications,
		  {N'EURO~'88}},
  pages		= {360--371},
  editor	= {L. Personnaz and G. Dreyfus},
  publisher	= {{EZIDET}},
  address	= {Paris},
  year		= 1989,
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buhmann92a,
  author	= {J. Buhmann and H. K{\"u}hnel},
  title		= {Complexity optimized vector quantization: a neural network
		  approach},
  booktitle	= {Proc. DCC '92, Data Compression Conf. },
  year		= {1992},
  editor	= {J. A. Storer and M. Cohn},
  pages		= {12--21},
  organization	= {IEEE; NASA/CESDIS},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA},
  x		= {The authors discuss a vector quantization strategy which
		  jointly optimizes distortion errors and complexity costs. .
		  . . Their approach establishes a unifying framework for
		  different quantization methods like . . . self-organizing
		  topological maps. },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buhmann92b,
  author	= {J. Buhmann and H. K{\"u}hnel},
  title		= {Unsupervised and Supervised Data Clustering with
		  Competitive Neural Networks},
  booktitle	= {Proc. IJCNN'92, International Conference on Neural
		  Networks},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  volume	= {IV},
  pages		= {796--801},
  year		= {1992},
  dbinsdate	= {oldtimer}
}

@Article{	  buhmann93a,
  author	= {J. Buhmann and H. K{\"u}hnel},
  journal	= {IEEE Trans. Information Theory},
  title		= {Vector Quantization with Complexity Costs},
  month		= {July},
  volume	= 39,
  number	= 4,
  year		= {1993},
  pages		= {1133--1145},
  dbinsdate	= {oldtimer}
}

@Article{	  buhmann93b,
  author	= {J. Buhmann and H. K{\"u}hnel},
  title		= {Complexity optimized data clustering by competitive neural
		  networks},
  journal	= {Neural Computation},
  year		= {1993},
  volume	= {5},
  number	= {1},
  pages		= {75--88},
  month		= {January},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buhusi93a,
  author	= {Catalin V. Buhusi},
  title		= {Neural Learning in Automatic Fuzzy Systems Synthesis},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {I},
  pages		= {786--789},
  organization	= {JNNS},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buhusi93b,
  author	= {Buhusi, C. V. },
  title		= {Parallel implementation of \mbox{self-organizing} neural
		  networks},
  booktitle	= {Romanian Symposium on Computer Science. 9th Symposium,
		  ROSYCS'93. Proceedings},
  year		= {1993},
  editor	= {Felea, V. and Ciobanu, G. },
  pages		= {51--8},
  organization	= {Institute for Comput. Sci. , Acad. of Sci. , Lasi,
		  Romania},
  publisher	= {Univ. Al. I. Cuza},
  address	= {Iasi, Romania},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  bunke01b,
  author	= {Bunke, H. and Gunter, S. and Jiang, X.},
  title		= {Towards bridging the gap between statistical and
		  structural pattern recognition: two new concepts in graph
		  matching},
  booktitle	= {Advances in Pattern Recognition---ICAPR 2001. Second
		  International Conference. Proceedings (Lecture Notes in
		  Computer Science Vol.2013). Springer-Verlag, Berlin,
		  Germany},
  year		= {2001},
  volume	= {},
  pages		= {1--11},
  abstract	= {Two concepts in structural pattern recognition are
		  discussed. The first, the median of a set of graphs, can be
		  used to characterize a set of graphs by just a single
		  prototype. Such a characterization is needed in various
		  tasks, for example, in clustering. The second concept is
		  the weighted mean of a pair of graphs. It can be used to
		  synthesize a graph that has a specified degree of
		  similarity, or distance, to each of a pair of given graphs.
		  Such an operation is needed in many machine-learning tasks.
		  It is argued that with these new concepts various
		  well-established techniques from statistical pattern
		  recognition become applicable in the structural domain,
		  particularly to graph representations. Specific examples
		  include k-means clustering, vector quantization, and
		  Kohonen maps.},
  dbinsdate	= {2002/1}
}

@Article{	  burel91a,
  author	= {G. Burel and I. Pottier},
  title		= {Vector quantization of images using {K}ohonen algorithm.
		  {T}heory and implementation},
  journal	= {Revue Technique Thomson-CSF},
  year		= {1991},
  volume	= {23},
  number	= {1},
  pages		= {137--159},
  month		= {March},
  dbinsdate	= {oldtimer}
}

@Article{	  burel92a,
  author	= {Gilles Burel},
  title		= {Nouveaux r{\'{e}}sultats th{\'{e}}oriques concernant les
		  cartes topologiques},
  journal	= {Bull. d'information des Laboratoires Centraux de Thomson
		  CSF},
  year		= {1992},
  publisher	= {Thomson CSF},
  address	= {Paris, France},
  volume	= {4},
  pages		= {3--13},
  note		= {(in french)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  burel93a,
  author	= {Gilles Burel and Jean-Yves Catros},
  title		= {Image Compression using Topological Maps and {MLP}},
  booktitle	= {Proc. ICNN'93, International Conference on Neural
		  Networks},
  year		= {1993},
  volume	= {II},
  pages		= {727--731},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  burel93b,
  author	= {Gilles Burel},
  title		= {Une nouvelle approche pour les r{\'{e}}seaux de neurones:
		  la repr{\'{e}}presentation scalaire Distribu{\'{e}}e},
  journal	= {Traitement du Signal},
  year		= {1993},
  volume	= {10},
  number	= {1},
  pages		= {41--51},
  note		= {(in french)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  burger97a,
  author	= {M. Burger and T. Graepel and K. Obermayer},
  title		= {Phase transitions in soft topographic vector
		  quantization},
  booktitle	= {Artificial Neural Networks---ICANN '97. 7th International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {W. Gerstner and A. Germond and M. Hasler and J. -D.
		  Nicoud},
  address	= {Berlin, Germany},
  pages		= {619--24},
  dbinsdate	= {oldtimer}
}

@InCollection{	  burger98a,
  author	= {M. Burger and T. Graepel and K. Obermayer},
  title		= {An Annealed Self-Organizing Map for Source Channel
		  Coding},
  booktitle	= {Advances in Neural Information Processing Systems},
  year		= {1998},
  editor	= {M. Jordan and M. Kearns and S. Solla},
  volume	= {10},
  pages		= {430--436},
  dbinsdate	= {oldtimer}
}

@Article{	  burke94a,
  author	= {Burke, L. I. },
  title		= {Neural methods for the traveling salesman problem:
		  insights from operations research},
  journal	= {Neural Networks},
  year		= {1994},
  volume	= {7},
  number	= {4},
  pages		= {681--90},
  abstract	= {Adaptive neural approaches to the traveling salesman
		  problem appear to hold great promise, as evidenced by the
		  work of Angeniol et al., among others. An adaptive approach
		  based on self-organizing feature maps with conscience---the
		  guilty net---has also exhibited the ability to give short
		  tour lengths for the TSP. Neural approaches rarely compare
		  in speed or efficacy, however, with good heuristics from
		  operations research. In this paper, experiments with the
		  guilty net and a new extension, the vigilant net, show that
		  adaptive approaches can benefit from established results in
		  operations research. Specifically, a simple starting
		  position strategy derives from operations research
		  literature, and proves an important force in reducing tour
		  lengths. In addition, operations research yields insights
		  into the kinds of TSP instances that should be studied.
		  Nonuniformly distributed cities pose an important problem,
		  and the guilty and vigilant nets yield tour lengths shorter
		  than even a comparable geometry preserving operations
		  research heuristic. Experimentation also shows the effect
		  of neighborhood size and conscience.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  burr88a,
  author	= {D. Burr},
  title		= {An improved elastic net method for the {T}ravelling
		  {S}alesman {P}roblem},
  booktitle	= {Proc. ICNN'88, International Conference on Neural
		  Networks},
  year		= {1988},
  volume	= {I},
  pages		= {69--76},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  burrascano90a,
  author	= {P. Burrascano and P. Lucci and G. Martinelli and R.
		  Perfetti},
  title		= {Velotopic maps in well-log inversion},
  booktitle	= {Proc. IJCNN-90, International Joint Conference on Neural
		  Networks, Washington, DC},
  year		= {1990},
  volume	= {I},
  pages		= {311--316},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  burrascano90b,
  author	= {P. Burrascano and P. Lucci and G. Martinelli and R.
		  Perfetti},
  title		= {Shear velocity estimation by the combined use of
		  supervised and unsupervised neural networks},
  booktitle	= {Proc. ICASSP-90, International Conference on Acoustics,
		  Speech and Signal Processing},
  year		= {1990},
  volume	= {IV},
  pages		= {1921--1924},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  x		= {A neural estimator composed of two different neural
		  networks, namely, a perceptron and a Kohonen map, is
		  proposed. . . . },
  dbinsdate	= {oldtimer}
}

@Article{	  burrascano91a,
  author	= {P. Burrascano},
  title		= {Learning vector quantization for the probabilistic neural
		  network},
  journal	= {IEEE Trans. on Neural Networks},
  year		= {1991},
  volume	= {2},
  number	= {4},
  pages		= {458--461},
  month		= {July},
  dbinsdate	= {oldtimer}
}

@Article{	  burrascano99a,
  author	= {Burrascano, P. and Fiori, S. and Mongiardo, M.},
  title		= {A review of artificial neural networks applications in
		  microwave computer-aided design},
  journal	= {International-Journal-of-RF-and-Microwave-Computer-Aided-Engineering}
		  ,
  year		= {1999},
  volume	= {9},
  pages		= {158--74},
  abstract	= {Neural networks found significant applications in
		  microwave CAD. In this paper, after providing a brief
		  description of neural networks employed so far in this
		  context, we illustrate some of their most significant
		  applications and typical issues arising in practical
		  implementation. We also summarize current research
		  tendencies and introduce use of self-organizing maps
		  enhancing model accuracy and applicability. We conclude
		  considering some future developments and exciting
		  perspectives opened from use of neural networks in
		  microwave CAD.},
  dbinsdate	= {2002/1}
}

@Article{	  burton99a,
  author	= {Burton, R.~M. and Plaehn, D.~C.},
  title		= {One-Dimensional {K}ohonen Maps Are Super-Stable with
		  Exponential Rate},
  journal	= {Advances in Applied Probability},
  year		= {1999},
  volume	= {31},
  number	= {2},
  pages		= {367--393},
  dbinsdate	= {oldtimer}
}

@Article{	  busch94a,
  author	= {C. Busch and M. H. Gross},
  title		= {Interactive Neural Network Texture Analysis and
		  Visualization for Surface Reconstruction in Medical
		  Imaging},
  journal	= {EUROGRAPHICS'93},
  year		= {1993},
  volume	= {12},
  number	= {3},
  pages		= {C-49--60},
  dbinsdate	= {oldtimer}
}

@Article{	  busch97a,
  author	= {C. Busch},
  title		= {Wavelet based texture segmentation of multi-modal
		  tomographic images},
  journal	= {Computers \& Graphics},
  year		= {1997},
  volume	= {21},
  number	= {3},
  pages		= {347--58},
  dbinsdate	= {oldtimer}
}

@InCollection{	  butchart95a,
  author	= {K. Butchart and N. Davey and R. Adams},
  title		= {A comparative study of three neural networks that use soft
		  competition},
  booktitle	= {From Natural to Artificial Neural Computation.
		  International Workshop on Artificial Neural Networks.
		  Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1995},
  editor	= {J. Mira and F. Sandoval},
  address	= {Berlin, Germany},
  pages		= {308--14},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  buttress95a,
  author	= {Buttress, J. and Frith, A. M. and Gent, C. R. and
		  Beaumont, A. J. },
  title		= {Using the {K}ohonen self organising map for novel data
		  handling in adaptive learning},
  booktitle	= {Neural Networks---Producing Dependable Systems (ERA
		  95--0973)},
  year		= {1995},
  pages		= {5. 1. 1--9},
  organization	= {EDS Defence Ltd. , UK},
  publisher	= {ERA Technol},
  address	= {Leatherhead, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  byrne94a,
  author	= {Byrne, W. and Mastrogiannis, K. and Meyer, G. F. },
  title		= {Classification of multi-spectral remote sensing data with
		  neural networks: a comparative study},
  booktitle	= {IEE Colloquium on 'Applications of Neural Networks to
		  Signal Processing' (Digest No. 1994/248)},
  year		= {1994},
  pages		= {5/1--2},
  organization	= {Dept. of Comput. Sci. , Keele Univ. , UK},
  publisher	= {IEE},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cabello93a,
  author	= {Cabello, D. and Penedo, M. G. and Barro, S. and Pardo, J.
		  M. and Heras, J. },
  title		= {CT image segmentation by \mbox{self-organizing} learning},
  booktitle	= {New Trends in Neural Computation. International Workshop
		  on Artificial Neural Networks. IWANN '93 Proceedings},
  year		= {1993},
  editor	= {Mira, J. and Cabestany, J. and Prieto, A. },
  pages		= {651--6},
  organization	= {Univ. de Santiago de Compostela, Spain},
  publisher	= {Springer-Verlag},
  address	= {Berlin, Germany},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cadieux00a,
  author	= {Cadieux, S. and Michaud, F. and Lalonde, F.},
  title		= {Intelligent system for automated fish sorting and
		  counting},
  booktitle	= {IEEE International Conference on Intelligent Robots and
		  Systems},
  year		= {2000},
  editor	= {},
  volume	= {2},
  pages		= {1279--1284},
  organization	= {Syst. d'automatismes et de mesure, Hydro-Quebec's Research
		  Cent. (IREQ)},
  publisher	= {},
  address	= {},
  abstract	= {This paper presents an automated system for counting fish
		  by species. This system is to be used in fishways for
		  monitoring and surveying fish. The system requires very few
		  adjustments and no special installation. An infrared
		  silhouette sensor is used to acquire the fish silhouettes.
		  These silhouettes are then processed on a personal computer
		  for fish counting and classification by species. The system
		  allows the operator to select the species of interest
		  according to the fauna of the specified river.
		  Classification is made based on the combined results of a
		  Bayes maximum likelihood classifier, a Learning Vector
		  Quantification classifier and a One-Class-One-Network
		  (OCON) neural network classifier. Through the use of
		  specialized classifiers of different types, a robust,
		  modular and expandable recognition system is created.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  cagnoni94a,
  author	= {Stefano Cagnoni and Guido Valli},
  title		= {{OS {LVQ} }: a training strategy for optimum-size Learning
		  Vector Quantization classifiers},
  pages		= {762--765},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {modification, multi-step training},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cagnoni97a,
  author	= {Cagnoni, A. and Mauri, G. and Pensini, M. P.},
  title		= {An hybrid architecture for on-line defects classification
		  in automatic inspection systems},
  booktitle	= {Neural Networks in Engineering Systems. Proceedings of the
		  1997 International Conference on Engineering Applications
		  of Neural Networks},
  publisher	= {Systems Engineering Association, Turku, Finland},
  year		= {1997},
  volume	= {1},
  pages		= {329--32},
  abstract	= {The paper describes a new general-purpose neural
		  architecture for an industrial quality control system,
		  BABINet. The architecture is especially tailored for defect
		  patterns whose parameters are detected by an automatic
		  inspection system, but it can be employed with success in
		  other similar classification problems. The BABINet system
		  is composed of two modules: an unsupervised module (whose
		  basic components are ART2 nets) and a supervised module
		  based on a new type of neural network model called ART2SUP.
		  Each module is composed of some networks arranged in
		  different ways: a tree structure for the unsupervised
		  module and a linear structure for the supervised one. The
		  unsupervised module creates groups of defect patterns which
		  are in detail classified by the supervised module.
		  Classification of the available defect patterns results in
		  a meaningful improvement (more than 10%) of the percentage
		  of correctness with respect to other studied neural models
		  (backpropagation, radial basis functions, learning vector
		  quantization) differently arranged in linear or complex
		  structures. At the moment the system is available on single
		  processor machines, but an implementation on parallel
		  machines is foreseen.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cai93a,
  author	= {S. Cai and H. Toral and J. Qiu},
  title		= {Flow Regime Identification by a Self-Organizing Neural
		  Network},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {868},
  publisher	= {Springer},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cai93b,
  author	= {Shiqian Cai and Haluk Toral},
  title		= {Flowrate Measurement in Air-Water Horizontal Pipeline by
		  Neural Network},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {II},
  pages		= {2013--2016},
  organization	= {JNNS},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {The Kohonen self-organising feature map (KSOFM) and the
		  multi-layer back propagation network (MBPN) were applied in
		  a hybrid network model to measure the flow rate of
		  individual phases in horizontal air-water flow. Feature
		  sets derived from turbulent absolute and differential
		  pressure signals obtained from a range of flow regimes were
		  classified into clusters by the KSOFM according to flow
		  regime. Samples belonging to each cluster were trained by
		  the MBPN to measure the flow rate of individual phases. Two
		  thirds of the samples were randomly selected to train the
		  MBPN, the remainder was used for testing. Individual phase
		  flow rates were identified with 10% accuracy.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cai94a,
  author	= {Yudong Cai},
  title		= {The Application of the Artificial Neural Network in the
		  Grading of Beer Quality},
  booktitle	= {Proc. WCNN'94, World Congress on Neural Networks},
  year		= {1994},
  volume	= {I},
  pages		= {516--520},
  organization	= {INNS},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  annote	= {application, pattern recognition},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  calinescu94a,
  author	= {Calinescu, R. and Grigoras, D. },
  title		= {A neural \mbox{self-organizing} scheme for dynamic load
		  allocation},
  booktitle	= {Transputer Applications and Systems'94. Proceedings of the
		  1994 World Transputer Congress},
  year		= {1994},
  editor	= {de Gloria, A. and Jane, M. R. and Marini, D. },
  pages		= {860--8},
  organization	= {Dept. of Comput. Sci. , Tech. Univ. Gh. Asachi, Iasi,
		  Romania},
  publisher	= {IOS Press},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@Article{	  callan99a,
  author	= {Callan, D. E. and Kent, R. D. and Roy, N. and Tasko, S.
		  M.},
  title		= {Self-organizing map for the classification of normal and
		  disordered female voices},
  journal	= {Journal of Speech, Language, and Hearing Research},
  year		= {1999},
  volume	= {42},
  pages		= {355--66},
  abstract	= {The goal of this research was to train a self-organizing
		  map (SOM) on various acoustic measures (amplitude
		  perturbation quotient, degree of voice breaks, rahmonic
		  amplitude, soft phonation index, standard deviation of the
		  fundamental frequency, and peak amplitude variation) of the
		  sustained vowel /a/ to enhance visualization of the
		  multidimensional nonlinear regularities inherent in the
		  input data space. The {SOM} was trained using 30 spasmodic
		  dysphonia exemplars, 30 pretreatment functional dysphonia
		  exemplars, 30 post-treatment functional dysphonia
		  exemplars, and 30 normal voice exemplars. After training,
		  the classification performance of the {SOM} was evaluated.
		  The results indicated that the {SOM} had better
		  classification performance than that of a stepwise
		  discriminant analysis over the original data. Analysis of
		  the weight values across the {SOM}, by means of stepwise
		  discriminant analysis, revealed the relative importance of
		  the acoustic measures in classification of the various
		  groups. The {SOM} provided both an easy way to visualize
		  multidimensional data, and enhanced statistical
		  predictability at distinguishing between the various groups
		  (over that conducted on the original data set). The authors
		  regard the results of this study as a promising initial
		  step into the use of {SOM}s with multiple acoustic measures
		  to assess phonatory function.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  calonge97a,
  author	= {T. Calonge and L. Alonso and R. Ralha and A. L. Sanchez},
  title		= {Parallel implementation of non-recurrent neural networks},
  booktitle	= {Vector and Parallel Processing---VECPAR '96. Second
		  International Conference on Vector and Parallel Processing
		  ---Systems and Applications. Selected Papers},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {J. M. L. M. Palma and J. Dongarra},
  address	= {Berlin, Germany},
  pages		= {314--25},
  dbinsdate	= {oldtimer}
}

@Article{	  camastra01a,
  author	= {Camastra, F. and Vinciarelli, A.},
  title		= {Cursive character recognition by learning vector
		  quantization},
  journal	= {Pattern Recognition Letters},
  year		= {2001},
  volume	= {22},
  number	= {6--7},
  month		= {May},
  pages		= {625--629},
  organization	= {Elsag spa},
  publisher	= {},
  address	= {},
  abstract	= {This paper presents a cursive character recognizer
		  embedded in an off-line cursive script recognition system.
		  The recognizer is composed of two modules: The first one is
		  a feature extractor, the second one a learning vector
		  quantizer. The selected feature set was compared to Zernike
		  polynomials using the same classifier. Experiments are
		  reported on a database of about 49,000 isolated characters.
		  },
  dbinsdate	= {2002/1}
}

@InCollection{	  cameron95a,
  author	= {B. M. Cameron and A. Manduca and R. A. Robb},
  title		= {Surface generation for virtual reality displays with a
		  limited polygonal budget},
  booktitle	= {Proceedings of the International Conference on Image
		  Processing},
  publisher	= {IEEE Computer Society Press},
  year		= {1995},
  volume	= {1},
  address	= {Los Alamitos, CA, USA},
  pages		= {438--41},
  dbinsdate	= {oldtimer}
}

@Article{	  cammarata89a,
  author	= {S. Cammarata},
  title		= {Introduction to neural computation},
  journal	= {Sistemi et Impresa},
  year		= {1989},
  volume	= {35},
  number	= {302},
  pages		= {688--697},
  month		= {April},
  note		= {(in Italian)},
  x		= { Outlines the historical development of neural network
		  approaches to computation. },
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cammarata93a,
  author	= {G. Cammarata and S. Cavalieri and A. Fichera and L.
		  Marletta},
  title		= {Self-Organizing Map to Filter Acoustic Mapping Survey in
		  Noise Pollutation Analysis},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {II},
  pages		= {2017--2020},
  organization	= {JNNS},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {In this paper the authors propose a neural approach to
		  filter the data provided by acoustic measurements. It is
		  based on the use of a Kohonen Self-Organizing Map network
		  which, in the learning phase receives correct acoustic
		  measurements. The Kohonen neural network learning on the
		  basis of this set of measurements would allow the network
		  to be used as a filter. Having received a set of acoustic
		  measurements in input, it would be able, in the production
		  phase, to discard any acoustic measurements which were
		  insignificant or affected by errors.},
  dbinsdate	= {oldtimer}
}

@Article{	  campanario95a,
  author	= {Juan Miguel Campanario},
  title		= {Using Neural Networks to Study Networks of Scientific
		  Journals},
  journal	= {Scientometrics},
  year		= {1995},
  volume	= {33},
  number	= {1},
  pages		= {23--40},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  campbell96a,
  author	= {Campbell, N. W. and Thomas, B. T. and Troscianko, T.},
  title		= {Neural networks for the segmentation of outdoor images},
  booktitle	= {Solving Engineering Problems with Neural Networks.
		  Proceedings of the International Conference on Engineering
		  Applications of Neural Networks (EANN'96). Syst. Eng.
		  Assoc, Turku, Finland},
  year		= {1996},
  volume	= {1},
  pages		= {343--6},
  abstract	= {One of the most crucial steps in many engineering
		  applications of computer vision is that of segmentation.
		  This is the process by which images are divided into
		  coherent regions based upon the colour, texture or other
		  properties of their pixels. This paper has two objectives.
		  The first of these is to demonstrate how a self-organising
		  feature map (SOM) may be used to automatically segment
		  images. The second aim is to examine the quality of these
		  segmentations when solely colour information is used, and
		  also when both colour and texture information is used. The
		  images used here are taken from the Bristol image database
		  and consist of a large number of high-quality colour
		  outdoor scenes. Every object in these scenes has been
		  assigned a label by hand. Hence a human-derived `ideal'
		  segmentation is available for each image, and hence a
		  quantitative performance measure for the segmentation
		  process is available.},
  dbinsdate	= {oldtimer}
}

@Article{	  campbell97a,
  author	= {N. W. Campbell and B. T. Thomas and T. Troscianko},
  title		= {Automatic segmentation and classification of outdoor
		  images using neural networks},
  journal	= {International Journal of Neural Systems},
  year		= {1997},
  volume	= {8},
  number	= {1},
  pages		= {137--44},
  dbinsdate	= {oldtimer}
}

@Article{	  campos00a,
  author	= {Campos, Marcos M. and Carpenter, Gail A.},
  title		= {Building adaptive basis functions with a continuous
		  \mbox{self-organizing} map},
  journal	= {Neural Processing Letters},
  year		= {2000},
  number	= {1},
  volume	= {11},
  pages		= {59--78},
  abstract	= {This paper introduces CSOM, a continuous version of the
		  Self-Organizing Map (SOM). The CSOM network generates maps
		  similar to those created with the original {SOM} algorithm
		  but, due to the continuous nature of the mapping, CSOM
		  outperforms the {SOM} on function approximation tasks. CSOM
		  integrates self-organization and smooth prediction into a
		  single process. This is a departure from previous work that
		  required two training phases, one to self-organize a map
		  using the {SOM} algorithm, and another to learn a smooth
		  approximation of a function. System performance is
		  illustrated with three examples.},
  dbinsdate	= {oldtimer}
}

@Article{	  campos01a,
  author	= {Campos, M. M. and Carpenter, G. A.},
  title		= {S-{TREE}: Self-organizing trees for data clustering and
		  online vector quantization},
  journal	= {Neural Networks},
  year		= {2001},
  volume	= {14},
  number	= {4--5},
  month		= {},
  pages		= {505--525},
  organization	= {Center for Adaptive Systems, Department Cognitive/Neural
		  Systems, Boston University},
  publisher	= {},
  address	= {},
  abstract	= {This paper introduces S-TREE (Self-Organizing Tree), a
		  family of models that use unsupervised learning to
		  construct hierarchical representations of data and online
		  tree-structured vector quantizers. The S-TREE1 model, which
		  features a new tree-building algorithm, can be implemented
		  with various cost functions. An alternative implementation,
		  S-TREE2, which uses a new double-path search procedure, is
		  also developed. The performance of the S-TREE algorithms is
		  illustrated with data clustering and vector quantization
		  examples, including a Gauss-Markov source benchmark and an
		  image compression application. S-TREE performance on these
		  tasks is compared with the standard tree-structured vector
		  quantizer (TSVQ) and the generalized Lloyd algorithm (GLA).
		  The image reconstruction quality with S-TREE2 approaches
		  that of GLA while taking less than 10% of computer time.
		  S-TREE1 and S-TREE2 also compare favorably with the
		  standard TSVQ in both the time needed to create the
		  codebook and the quality of image reconstruction. Copyright
		  },
  dbinsdate	= {2002/1}
}

@Article{	  campos96a,
  author	= {T. P. R. Campos},
  title		= {Connectionist modeling for arm kinematics using visual
		  information},
  journal	= {IEEE Transactions on Systems, Man and Cybernetics, Part B
		  [Cybernetics]},
  year		= {1996},
  volume	= {26},
  number	= {1},
  pages		= {89--99},
  dbinsdate	= {oldtimer}
}

@InCollection{	  campos98a,
  author	= {M. M. Campos and G. A. Carpenter},
  title		= {{WSOM}: building adaptive wavelets with
		  \mbox{self-organizing} maps},
  booktitle	= {1998 IEEE International Joint Conference on Neural
		  Networks Proceedings. IEEE World Congress on Computational
		  Intelligence},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {1},
  address	= {New York, NY, USA},
  pages		= {763--7},
  dbinsdate	= {oldtimer}
}

@InCollection{	  campos98b,
  author	= {Marcos M. Campos and Gail A. Carpenter},
  title		= {Building Adaptive Basis Functions with a Continuous
		  {SOM}},
  booktitle	= {Proc. JCIS'98},
  publisher	= {Association for Intelligent Machinery, Inc},
  year		= 1998,
  editor	= {Paul P. Wang},
  volume	= {II},
  pages		= {68--71},
  abstract	= {This paper introduces CSOM, a distributed version of the
		  self-organizing map network capable of generating maps
		  similar to those created with the original algorithm: due
		  to the continuous nature of the mapping, CSOM outperforms
		  the traditional {SOM} algorithm in function approximation
		  tasks. System performance is illustrated with three
		  examples.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  canas91a,
  author	= {A. Canas and J. Ortega and F. J. Fernandez and A. Prieto
		  and F. J. Pelayo},
  title		= {An approach to isolated word recognition using multilayer
		  perceptrons},
  booktitle	= {Proc. IWANN'91, Int. Workshop on Artificial Neural
		  Networks},
  year		= {1991},
  editor	= {A. Prieto},
  pages		= {340--347},
  publisher	= {Springer},
  address	= {Berlin, Heidelberg},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cao95a,
  author	= {Yuanda Cao and Yifeng Chen},
  title		= {A neural Spatio-Temporal Feature Detector},
  booktitle	= {Proceedings of International Conference on Neural
		  Information Processing (ICONIP `95)},
  year		= {1995},
  editor	= {Zhong, Y. and Yang, Y. and Wang, M. },
  volume	= {1},
  pages		= {201--4},
  organization	= {Inst. of Artificial Intelligence, Beijing Inst. of
		  Technol. , China},
  publisher	= {Publishing House of Electron. Ind},
  address	= {Beijing, China},
  dbinsdate	= {oldtimer}
}

@Article{	  capitan-vallvey00a,
  author	= {Capitan-Vallvey, L. F. and Navas, N. and del Olmo, M. and
		  Consonni, V. and Todeschini, R.},
  title		= {Resolution of mixtures of three nonsteroidal
		  anti-inflammatory drugs by fluorescence using partial least
		  squares multivariate calibration with previous wavelength
		  selection by Kohonen artificial neural networks},
  journal	= {Talanta},
  year		= {2000},
  volume	= {52},
  number	= {6},
  month		= {Sep},
  pages		= {1069--1079},
  organization	= {Univ of Granada},
  publisher	= {Elsevier Science Publ Co Inc},
  address	= {New York, NY},
  abstract	= {A spectrofluorometric method for the quantitative
		  determination of flufenamic, mefenamic and meclofenamic
		  acids in mixtures has been developed by recording emission
		  fluorescence spectra between 370 and 550 nm with an
		  excitation wavelength of 352 nm. The excitation-emission
		  spectra of these compounds are deeply overlapped which does
		  not allow their direct determination without previous
		  separation. The proposed method applies partial least
		  squares (PLS) multivariate calibration to the resolution of
		  this mixture using a set of wavelengths previously selected
		  by Kohonen artificial neural networks (K-ANN). The linear
		  calibration graphs used to construct the calibration matrix
		  were selected in the ranges from 0.25 to 1.00 \mu{}g
		  ml<sup>-1</sup> for flufenamic and meclofenamic acids, and
		  from 1.00 to 4.00 \mu{}g ml<sup>-1</sup> for mefenamic
		  acid. A cross-validation procedure was used to select the
		  number of factors. The selected calibration model has been
		  applied to the determination of these compounds in
		  synthetic mixtures and pharmaceutical formulations.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  cappelli95a,
  author	= {Marco Cappelli and Rodolfo Zunino},
  title		= {{D {LVQ} }: Dynamic model for {L}earning {V}ector
		  {Q}uantization},
  volume	= {I},
  pages		= {652--655},
  booktitle	= {Proc. WCNN'95, World Congress on Neural Networks},
  year		= {1995},
  organization	= {INNS},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  caravaglia01a,
  author	= {Susan Caravaglia},
  title		= {Statistical Analysis of the {T}animoto Coefficient
		  Self-Organizing Map ({TCSOM}) Applied to Health Behavioral
		  Survey Data},
  booktitle	= {International Joint Conference on Neural Networks,
		  Washington, DC, USA, July 15--19},
  pages		= {},
  year		= {2001},
  month		= {July},
  note		= {CD-ROM},
  dbinsdate	= {2002/1}
}

@InProceedings{	  card95a,
  author	= {H. C. Card and SriGouri Kamarsu},
  title		= {Limited Precision Unsupervised Learning Algorithms for
		  Speech Coding},
  volume	= {I},
  pages		= {128--131},
  booktitle	= {Proc. WCNN'95, World Congress on Neural Networks},
  year		= {1995},
  organization	= {INNS},
  dbinsdate	= {oldtimer}
}

@Book{		  cardei97a,
  author	= {Cardei, V. C. and Hadley, R. F.},
  title		= {Predicting semantic categories using a hybrid neural
		  network. Technical report no. CSS-IS TR97--02.},
  year		= {1997},
  abstract	= {Describes a hybrid neural network that can predict the
		  semantic categories and some characteristic features of the
		  words from a given sentence. Section 1 presents the general
		  description of the neural network and the way it functions,
		  and the considerations which led to the network
		  architecture. Section 2 describes the Kohonen map and
		  output network, along with their corresponding learning
		  algorithms in the form of a pseudo-code. Section 3
		  describes the structure of the program written for
		  implementing the hybrid network and discusses a sample of a
		  test run. The final section presents some experimental data
		  and comparisons between average activations for nouns in
		  novel positions versus the nouns from the training set.
		  Other data that indicates the degree of systematicity of
		  the network are also presented.},
  dbinsdate	= {oldtimer}
}

@Article{	  cardiel01a,
  author	= {Cardiel, G. B. and {De los Angeles Fabian Alvarez}, M. and
		  Martinez, M. V. and Villasenor, L.},
  title		= {Inteligencias artificiales y ensayos ultrasonicos para la
		  deteccion de defectos},
  journal	= {Revista de Metalurgia (Madrid)},
  year		= {2001},
  volume	= {37},
  number	= {3},
  month		= {May/June },
  pages		= {403--411},
  organization	= {Instituto de Invest. Metalurgicas, Universidad
		  Michoacana},
  publisher	= {},
  address	= {},
  abstract	= {One of the most serious problems in the quality control of
		  welded unions is the location, identification and
		  classification of defects. As a solution to this problem, a
		  technique for classification, applicable to welded unions
		  done by electric arc welding as well as by friction, is
		  proposed; it is based on ultrasonic signals. The neuronal
		  networks proposed are Kohonen and Multilayer Perceptron,
		  all in a virtual instrument environment. Currently the
		  techniques most used in this field are: radiological
		  analysis (X-rays) and ultrasonic analysis (ultrasonic
		  waves). The X-ray technique in addition to being dangerous
		  requires highly specialized personnel and equipment,
		  therefore its use is restricted. The ultrasonic technique,
		  in spite of being one of the most used for detection of
		  discontinuities, requires personnel with wide experience in
		  the interpretation of ultrasonic signals; this is a
		  time-consuming process which necessarily increases its
		  operation cost. The classification techniques that we
		  propose turn out to be safe, reliable, inexpensive and easy
		  to implement for the solution of this important problem.},
  dbinsdate	= {2002/1}
}

@Article{	  carl94a,
  author	= {Carl, S. and Kraft, R. },
  title		= {Land use classification of {ERS-1} images with an
		  artificial neural network},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1994},
  volume	= {2315},
  pages		= {452--9},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carlen93a,
  author	= {Carlen, E. T. and Abdel-Aty-Zohdy, H. S. },
  title		= {{VLSI} implementation of a feature mapping neural
		  network},
  booktitle	= {Proceedings of the 36th Midwest Symposium on Circuits and
		  Systems},
  year		= {1993},
  volume	= {2},
  pages		= {958--62},
  organization	= {Dept. of Electr. \& Syst. Eng. , Oakland Univ. ,
		  Rochester, MI, USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  carlini97a,
  author	= {P. Carlini},
  title		= {Self organizing maps, vector quantization, and fractal
		  image coding},
  journal	= {Fractals},
  year		= {1997},
  volume	= {5},
  number	= {suppl. issue},
  pages		= {201--14},
  note		= {(Fractal Image Encoding and Analysis Conf. Date: 8--17
		  July 1995 Conf. Loc: Trondheim, Norway Conf. Sponsor:
		  NATO)},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carlson91a,
  author	= {Eero Carlson},
  title		= {Self-Organizing Feature Maps for Appraisal of Land Value
		  of Shore Parcels},
  booktitle	= {Artificial Neural Networks},
  year		= {1991},
  editor	= {T. Kohonen and K. M{\"{a}}kisara and O. Simula and J.
		  Kangas},
  volume	= {II},
  pages		= {1309--1312},
  publisher	= {North-Holland},
  address	= {Amsterdam, Netherlands},
  month		= {},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carlson93a,
  author	= {Eero Carlson},
  title		= {Cognitive Grammar and Map Digitization},
  booktitle	= {Proc. ICANN'93, International Conference on Artificial
		  Neural Networks},
  year		= {1993},
  editor	= {Stan Gielen and Bert Kappen},
  pages		= {1018},
  publisher	= {Springer},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InCollection{	  carlson97a,
  author	= {Eero Carlson},
  title		= {Scaling and sensitivity in appraisal},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {57--62},
  dbinsdate	= {oldtimer}
}

@InCollection{	  carlson98a,
  author	= {E. Carlson},
  title		= {Real Estate Investment Appraisal of Land Properties using
		  {SOM}},
  booktitle	= {Visual Explorations in Finance with Self-Organizing Maps},
  publisher	= {Springer},
  year		= {1998},
  editor	= {G. Deboeck and T. Kohonen},
  address	= {London},
  pages		= {117--127},
  dbinsdate	= {oldtimer}
}

@Article{	  carosone95a,
  author	= {F. Carosone and A. Cenedese and G. Querzoli},
  title		= {Recognition of partially overlapped particle images using
		  the {K}ohonen neural network},
  journal	= {Experiments in Fluids},
  year		= {1995},
  volume	= {19},
  number	= {4},
  pages		= {225--232},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carpinteiro00a,
  author	= {Carpinteiro, Otavio A. S. and {da Silva}, Alexandre P. A.
		  and Feichas, Carlos H. L.},
  title		= {Hierarchical neural model in short-term load forecasting},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2000},
  editor	= {},
  volume	= {6},
  pages		= {241--246},
  organization	= {Escola Federal de Engenharia de Itajuba},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {This paper proposes a novel neural model to the problem of
		  short-term load forecasting. The neural model is made up of
		  two self-organizing map nets---one on top of the other. It
		  has been successfully applied to domains in which the
		  context information given by former events plays a primary
		  role. The model was trained and assessed on load data
		  extracted from a Brazilian electric utility. It was
		  required to predict once every hour the electric load
		  during the next 24 hours. The paper presents the results,
		  and evaluates them.},
  dbinsdate	= {2002/1}
}

@Article{	  carpinteiro00b,
  author	= {Carpinteiro, O. A. S.},
  title		= {A hierarchical self-organising map model for sequence
		  recognition},
  journal	= {PATTERN ANALYSIS AND APPLICATIONS},
  year		= {2000},
  volume	= {3},
  number	= {3},
  pages		= {279--287},
  abstract	= {This raper presents art analysis of an original
		  hierarchical neural model on a complex sequence---che
		  complete sixteenth four-part fugue in G minor of the
		  Well-Tempered Clavier (vol. I) of J. S. Bach. The model
		  makes an effective use of context information, through in
		  hierarchical topology and its embedded time integrators,
		  and that enables it to keep a very good account of
		  patterns. The model peri;,rms sequence classification and
		  discrimination efficiently. It has application in domains
		  which require pattern recognition, or particularly, which
		  demand recognising either a set of sequences of vectors in
		  time, or sub-sequences into a unique and large sequence of
		  vectors in time.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  carpinteiro00c,
  author	= {Carpinteiro, O. A. S. and {Alves da Silva} A. P.},
  title		= {A hierarchical neural model in short-term load
		  forecasting},
  booktitle	= {Proceedings. Vol.1. Sixth Brazilian Symposium on Neural
		  Networks. IEEE Comput. Soc, Los Alamitos, CA, USA},
  year		= {2000},
  volume	= {},
  pages		= {120--4},
  abstract	= {This paper proposes a novel neural model for the
		  short-term load forecasting problem. The neural model is
		  made up of two self-organizing map nets-one on top of the
		  other. It has been successfully applied to domains in which
		  the context information given by former events plays a
		  primary role. The model was trained and assessed on the
		  load data extracted from a Brazilian electric utility. It
		  was required to predict once every hour the electric load
		  during the next 24 hours. The paper presents the results
		  and evaluates them.},
  dbinsdate	= {2002/1}
}

@Article{	  carpinteiro01a,
  author	= {Carpinteiro, O. A. S. and {Alves Da Silva}, A. P.},
  title		= {A hierarchical self-organizing map model in short-term
		  load forecasting},
  journal	= {Journal of Intelligent and Robotic Systems: Theory and
		  Applications},
  year		= {2001},
  volume	= {31},
  number	= {1--3},
  month		= {May/July },
  pages		= {105--113},
  organization	= {Instituto de Engenharia Eletrica, Escola Fed. de
		  Engenharia de Itajuba},
  publisher	= {},
  address	= {},
  abstract	= {This paper proposes a novel neural model to the problem of
		  short-term load forecasting. The neural model is made up of
		  two self-organizing map nets---one on top of the other. It
		  has been successfully applied to domains in which the
		  context information given by former events plays a primary
		  role. The model was trained and assessed on load data
		  extracted from a Brazilian electric utility. It was
		  required to predict once every hour the electric load
		  during the next 24 hours. The paper presents the results,
		  and evaluates them.},
  dbinsdate	= {2002/1}
}

@InCollection{	  carpinteiro98a,
  author	= {O. A. S. Carpinteiro},
  title		= {A \mbox{self-organizing} map model for analysis of musical
		  time series},
  booktitle	= {Proceedings 5th Brazilian Symposium on Neural Networks},
  publisher	= {IEEE Computer Society},
  year		= {1998},
  editor	= {A. de Padua Braga and T. B. Ludermir},
  address	= {Los Alamitos, CA, USA},
  pages		= {140--5},
  dbinsdate	= {oldtimer}
}

@InCollection{	  carpinteiro98b,
  author	= {Ot{\'a}vio Augusto S. Carpinteiro},
  title		= {A Hierarchical Self-Organizing Map Model for Sequence
		  Recognition},
  booktitle	= {Proceedings of ICANN98, the 8th International Conference
		  on Artificial Neural Networks},
  publisher	= {Springer},
  year		= 1998,
  editor	= {L. Niklasson and M. Bod{\'e}n and T. Ziemke},
  volume	= 2,
  address	= {London},
  pages		= {815--820},
  abstract	= {Proposes a neural model made up of two self-organizing map
		  nets-one on top of the other. The model makes an effective
		  use of context information, and that enables it to perform
		  sequence classification and discrimination efficiently. It
		  was trained and assessed on a four-part fugue of J.S. Bach.
		  The model has application in domains which require pattern
		  recognition, or in particular, which demand recognizing
		  either a set of sequences of vectors in time or
		  sub-sequences into a unique and large sequence of vectors
		  in time.},
  dbinsdate	= {oldtimer}
}

@Article{	  carpinteiro99a,
  author	= {Carpinteiro, Otavio Augusto S.},
  title		= {Hierarchical \mbox{self-organizing} map model for sequence
		  recognition},
  journal	= {Neural Processing Letters},
  year		= {1999},
  number	= {3},
  volume	= {9},
  pages		= {209--220},
  abstract	= {A novel neural model made up of two self-organizing maps
		  nets---one on top of the other---is introduced and analyzed
		  experimentally. The model makes effective use of context
		  information, and that enables it to perform sequence
		  classification and discrimination efficiently. It was
		  successfully applied to real sequences, taken from the
		  third voice of the sixteenth four-part fugue in G minor of
		  the Well-Tempered Clavier (vol. I) of J.S. Bach. The model
		  has an application in domains which require pattern
		  recognition, or more specifically, which demand the
		  recognition of either a set of sequences of vectors in time
		  or sub-sequences into a unique and large sequence of
		  vectors in time.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carpinteiro99b,
  author	= {Carpinteiro, O. A. S. and {Alves da Silva}, A. P.},
  title		= {A hierarchical \mbox{self-organizing} map model in
		  short-term load forecasting},
  booktitle	= {Engineering Applications of Neural Networks. Proceedings
		  of the 5th International Conference on Engineering
		  Applications of Neural Networks (EANN'99)},
  publisher	= {Wydawnictwo Adam Marszalek},
  address	= {Torun, Poland},
  year		= {1999},
  volume	= {},
  pages		= {75--80},
  abstract	= {This paper proposes a novel neural model to the problem of
		  short-term load forecasting. The neural model is made up of
		  two self-organizing map nets-one on top of the other. It
		  has been successfully applied to domains in which the
		  context information given by former events plays a primary
		  role. The model was trained and assessed on load data
		  extracted from a Brazilian electric utility. It was
		  required to predict once every hour the electric load
		  during the next 24 hours. The paper presents the results,
		  and evaluates them.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carraro89a,
  author	= {A. Carraro and E. Chilton and H. McGurk},
  title		= {A telephonic lipreading device for the hearing impaired},
  booktitle	= {IEE Colloquium on 'Biomedical Applications of Digital
		  Signal Processing' (Digest No. 144)},
  year		= {1989},
  organization	= {IEE},
  publisher	= {IEE},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carrato93a,
  author	= {Sergio Carrato and Giovanni L. Sicuranza and Luigi Manzo},
  title		= {Application of Ordered Codebooks to Image Coding},
  booktitle	= {Neural Networks for Signal Processing 3---Proceedings of
		  the 1993 IEEE Workshop},
  year		= {1993},
  editor	= {Kamm, C. A. and Kung, S. Y. and Yoon, B. and Chellappa, R.
		  and Kung, S. Y. },
  pages		= {291--300},
  organization	= {IEEE},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, New Jersey, USA},
  month		= {September},
  dbinsdate	= {oldtimer}
}

@Article{	  carrato94a,
  author	= {Sergio Carrato},
  title		= {Image vector quantization using ordered codebooks:
		  Properties and applications},
  journal	= {Signal Processing},
  year		= {1994},
  volume	= {40},
  number	= {1},
  pages		= {87--103},
  annote	= {vector quantization, image compression},
  abstract	= {We consider the application of self-organizing feature
		  maps (SOFMs) to the problem of image vector quantization.
		  We show that, using a SOFM-based codebook, vector
		  quantizers which are better than conventionally designed
		  ones can be obtained if their property of ordering is
		  suitably exploited. In fact, there are at least three
		  possible applications, which are described in the paper,
		  where they give better results at no extra cost. We first
		  introduce a measure of disorder, which is used for
		  evaluating the quality of the ordering of SOFM-based
		  codebooks. Then, we exploit the ordering of these codebooks
		  for two purposes: a differential coding technique which
		  relies on the correlation between successive addresses
		  related to adjacent blocks, and a fast search algorithm
		  based on a `first guess' of the winning codeword. We also
		  show that an ordered codebook is robust with respect to
		  channel errors. Simulation results are presented, which
		  confirm the useful properties of SOFM-based vector
		  quantizers.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  carreira00a,
  author	= {Carreira, M. J. and Mirmehdi, M. and Thomas, B. T. and
		  Haddon, J. F.},
  title		= {Grouping of directional features using an extended Hough
		  transform},
  booktitle	= {Proceedings 15th International Conference on Pattern
		  Recognition. ICPR-2000. IEEE Comput. Soc, Los Alamitos, CA,
		  USA},
  year		= {2000},
  volume	= {3},
  pages		= {990--3},
  abstract	= {Directional features extracted from Gabor responses are
		  used as primitives for perceptual grouping. In previous
		  work, we extracted Gabor features in 8 directions and then
		  applied two self-organising maps, thus classifying each
		  pixel in the image within a neuron-map, each corner of
		  which represents one of four main directions. In this work
		  we group pixels with similar directional features to detect
		  salient structures within an image. Results obtained from
		  application to forward-looking infrared (FLIR) images are
		  very promising.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  carter93a,
  author	= {S. Carter and R. J. Frank and D. S. W. Tansley},
  title		= {Clone Detection in Telecommunications Software Systems: A
		  Neural Net Approach},
  booktitle	= {Proc. Int. Workshop on Application of Neural Networks to
		  Telecommunications},
  year		= {1993},
  editor	= {Joshua Alspector and Rodney Goodman and Timothy X. Brown},
  pages		= {273--287},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  annote	= {application, classification, symbolic data},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  caselli94a,
  author	= {Caselli, S. and Faldella, E. and Fringuelli, B. and Rosi,
		  L. },
  title		= {A neural approach to robotic haptic recognition of {3-D}
		  objects based on a {K}ohonen \mbox{self-organizing} feature
		  map},
  booktitle	= {IECON '94. 20th International Conference on Industrial
		  Electronics, Control and Instrumentation},
  year		= {1994},
  volume	= {2},
  pages		= {835--40},
  organization	= {Dipartimento di Ingegneria dell'Inf. , Parma Univ. ,
		  Italy},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  abstract	= {The paper describes a versatile robotic haptic recognition
		  system of three-dimensional objects. The adopted design
		  methodology foresees a learning phase of the geometric
		  properties of the objects, followed by the operative phase
		  of actual recognition in which the robot explores the
		  objects with its end-effector, correlating the sensorial
		  data with the preceding perceptive experiences. These
		  phases are mapped on the training and classification
		  activities typical of the unsupervised Kohonen neural
		  networks. The validity of the novel approach pursued for
		  the design of the haptic recognition system has been
		  ascertained with reference to a high-dexterity 3-finger,
		  10-degree of freedom robotic hand (the University of
		  Bologna hand), but the underlying methodological issues can
		  be specialized to any dextrous end-effector. The developed
		  prototype system, even though currently referring to a
		  simulated environment and to some working assumptions, such
		  as the convexity of the objects and their immobility during
		  exploration, has already shown a satisfactory operative
		  level in recognizing objects belonging to a set of
		  significant cardinality.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  castellano00a,
  author	= {Castellano, G. and Fanelli, A. M.},
  title		= {Self-organizing neural fuzzy inference network},
  booktitle	= {Proceedings of the International Joint Conference on
		  Neural Networks},
  year		= {2000},
  editor	= {},
  volume	= {5},
  pages		= {14--19},
  organization	= {Universita degli Studi di Bari},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {A self-organizing neural network is proposed which is
		  inherently a fuzzy inference system with the capability of
		  learning fuzzy rules from data. The learning strategy
		  consists of two phases: a self-organizing clustering to
		  establish the structure of the network as well as the
		  initial values of its parameters and a supervised learning
		  phase for optimal adjustment of these parameters. After
		  learning, the network encodes in its structure the
		  essential design parameters of a fuzzy system. An example
		  is given to illustrate the characteristics and capabilities
		  of the proposed network.},
  dbinsdate	= {2002/1}
}

@Article{	  castillo00a,
  author	= {Castillo, P. A. and Merelo, J. J. and Prieto, A. and
		  Rivas, V. and Romero, G.},
  title		= {G-Prop: Global optimization of multilayer perceptrons
		  using {GA}s},
  journal	= {Neurocomputing},
  year		= {2000},
  volume	= {35},
  number	= {},
  month		= {Nov},
  pages		= {149--163},
  organization	= {Universidad de Granada},
  publisher	= {Elsevier Science B.V.},
  address	= {Amsterdam},
  abstract	= {A general problem in model selection is to obtain the
		  right parameters that make a model fit observed data. For a
		  multilayer perceptron (MLP) trained with back-propagation
		  (BP), this means finding appropriate layer size and initial
		  weights. This paper proposes a method (G-Prop, genetic
		  backpropagation) that attempts to solve that problem by
		  combining a genetic algorithm (GA) and BP to train MLPs
		  with a single hidden layer. The GA selects the initial
		  weights and changes the number of neurons in the hidden
		  layer through the application of specific genetic
		  operators. G-Prop combines the advantages of the global
		  search performed by the GA over the MLP parameter space and
		  the local search of the BP algorithm. The application of
		  the G-Prop algorithm to several real-world and benchmark
		  problems shows that MLPs evolved using G-Prop are smaller
		  and achieve a higher level of generalization than other
		  perceptron training algorithms, such as Quick-Propagation
		  or RPROP, and other evolutive algorithms, such as G-LVQ.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  castillo99a,
  author	= {Castillo, P. A. and Merelo, J. J. and Gonzalez, J. and
		  Rivas, V. and Romero, G.},
  title		= {{SA}-{P}rop: optimization of multilayer perceptron
		  parameters using simulated annealing},
  booktitle	= {Foundations and Tools for Neural Modeling. International
		  Work-Conference on Artificial and Natural Neural Networks,
		  IWANN'99. Proceedings, (Lecture Notes in Computer Science
		  Vol.1606)},
  publisher	= {Springer-Verlag},
  address	= {Berlin, Germany},
  year		= {1999},
  volume	= {1},
  pages		= {661--70},
  abstract	= {A general problem in model selection is to obtain the
		  right parameters that make a model fit observed data. If
		  the model selected is a multilayer perceptron (MLP) trained
		  with backpropagation (BP), it is necessary to find
		  appropriate initial weights and learning parameters. This
		  paper proposes a method that combines simulated annealing
		  (SA) and BP to train MLPs with a single hidden layer,
		  termed SA-Prop. SA selects the initial weights and the
		  learning rate of the network. SA-Prop combines the
		  advantages of the stochastic search performed by the SA
		  over the MLP parameter space and the local search of the BP
		  algorithm. The application of the proposed method to
		  several real-world benchmark problems shows that MLPs
		  evolved using SA-Prop achieve a higher level of
		  generalization than other perceptron training algorithms,
		  such as QuickPropagation (QP) or RPROP, and other evolutive
		  algorithms, such as G-LVQ.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  castillo99b,
  author	= {Castillo, P. A. and Rivas, V. and Merelo, J. J. and
		  Gonzalez, J. and Prieto, A. and Romero, G.},
  title		= {{G}-{P}rop-{II}: global optimization of multilayer
		  perceptrons using {GA}s},
  booktitle	= {Proceedings of the 1999 Congress on Evolutionary
		  Computation---CEC99. },
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {},
  pages		= {2022--7},
  abstract	= {A general problem in model selection is to obtain the
		  right parameters that make a model fit observed data. For a
		  multilayer perceptron (MLP) trained with backpropagation
		  (BP), this means finding the right hidden layer size,
		  appropriate initial weights and learning parameters. The
		  paper proposes a method (G-Prop-II) that attempts to solve
		  that problem by combining a genetic algorithm (GA) and BP
		  to train MLPs with a single hidden layer. The GA selects
		  the initial weights and the learning rate of the network,
		  and changes the number of neurons in the hidden layer
		  through the application of specific genetic operators.
		  G-Prop-II combines the advantages of the global search
		  performed by the GA over the MLP parameter space and the
		  local search of the BP algorithm. The application of the
		  G-Prop-II algorithm to several real world and benchmark
		  problems shows that MLPs evolved using G-Prop-II are
		  smaller and achieve a higher level of generalization than
		  other perceptron training algorithms, such as
		  QuickPropagation or RPROP, and other evolutive algorithms,
		  such as G-LVQ. It also shows some improvement over previous
		  versions of the algorithm.},
  dbinsdate	= {oldtimer}
}

@Article{	  catrina93a,
  author	= {Catrina, S. },
  title		= {Nested network method for robot control},
  journal	= {Revue Roumaine des Sciences Techniques, Serie
		  Electrotechnique et Energetique},
  year		= {1993},
  volume	= {38},
  number	= {3},
  pages		= {421--8},
  month		= {July-Sept},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  caudill88a,
  author	= {M. Caudill},
  title		= {Network paradigm selection guidelines for application
		  development},
  booktitle	= {Proc. Fourth Annual Artificial Intelligence and Advanced
		  Computer Technology Conference},
  year		= {1988},
  pages		= {298--302},
  publisher	= {Tower Conf. Management},
  address	= {Glen Ellyn, IL},
  dbinsdate	= {oldtimer}
}

@Article{	  caudill93a,
  author	= {M. Caudill},
  title		= {A little knowledge is a dangerous thing (neural nets)},
  journal	= {AI Expert},
  year		= {1993},
  volume	= {8},
  number	= {6},
  pages		= {16--22},
  month		= {June},
  annote	= {A review of {SOFM}},
  dbinsdate	= {oldtimer}
}

@Article{	  cavazos00a,
  author	= {Cavazos, Tereza},
  title		= {Using self-organizing maps to investigate extreme climate
		  events: An application to wintertime precipitation in the
		  Balkans},
  journal	= {Journal of Climate},
  year		= {2000},
  volume	= {13},
  number	= {10},
  month		= {},
  pages		= {1718--1732},
  organization	= {Pennsylvania State Univ},
  publisher	= {American Meteorological Soc},
  address	= {Boston, MA},
  abstract	= {This paper examines some of the physical mechanisms and
		  remote linkages associated with extreme wintertime
		  precipitation in the Balkans. The analysis is assessed on
		  daily timescales to determine the role of the circulation
		  and atmospheric moisture on extreme events, and also at
		  intraseasonal and interannual timescales to find possible
		  linkages with the North Atlantic Oscillation (NAO) and the
		  Arctic Oscillation (AO) patterns. A nonlinear
		  classification known as the self-organizing map (SOM) is
		  employed to obtain the climate modes and anomalies that
		  dominated during the 1980--93 period. An artificial neural
		  network (ANN) is also used to derive daily precipitation at
		  gridpoint scale and at local scale in Bucharest, Romania.
		  Of the predictors used, 500--1000-hPa thickness, 700-hPa
		  geopotential heights, and 700-hPa moisture are the most
		  important controls of daily precipitation. These results
		  are substantiated with the climate states from the SOM
		  classification, which show strong meridional flow over
		  central and eastern Europe coupled to increased winter
		  disturbances in the central Mediterranean and a tongue of
		  moisture at the 700-hPa level from the eastern
		  Mediterranean and the Black Sea during anomalously wet
		  events in the Bulgarian region. Dry events are almost an
		  inverse of these conditions. Extreme events are further
		  modulated by changes in the circulation associated with the
		  AO. In contrast, the NAO does not play a role on wintertime
		  precipitation over the region. The ANN captures well
		  synoptic events and dry spells, but tends to overestimate
		  (underestimate) small (large) events. This suggests a
		  problem for area-averaged precipitation, which is already
		  biased by its spatial resolution. However, comparison
		  between precipitation at Bucharest station and at its
		  nearest grid point shows that the performance of the ANN is
		  slightly better at gridpoint scale.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  caviglia90a,
  author	= {D. D. Caviglia and G. M. Bisio and F. Curatelli and L.
		  Giovannacci and L. Raffo},
  title		= {Pre-placement of {VLSI} blocks through learning neural
		  networks},
  booktitle	= {Proc. EDAC, European Design Automation Conf. , Glasgow,
		  Scotland},
  year		= {1990},
  pages		= {650--654},
  organization	= {IEEE; EDAC},
  publisher	= {IEEE Computer Society Press},
  address	= {Washington, DC},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cawley93a,
  author	= {G. C. Cawley and P. D. Noakes},
  title		= {The Use of Vector Quantization in Neural Speech
		  Synthesis},
  booktitle	= {Proc. IJCNN-93, International Joint Conference on Neural
		  Networks, Nagoya},
  year		= {1993},
  volume	= {III},
  pages		= {2227--2230},
  organization	= {JNNS},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {Our previous work has indicated that multi-layer
		  perceptrons (MLPs) trained using the backpropagation (BP)
		  algorithm, have great difficulty in learning continuous
		  mappings with sufficient accuracy for speech synthesis. The
		  use of vector quantization allows networks to be trained to
		  select a sequence of entries from a codebook of speech
		  parameter vectors. For the network to be able to generalize
		  meaningfully some correlation must exist between codebook
		  vectors and the indices by which they are recalled
		  (otherwise the network will be attempting to learn an
		  essentially random mapping). This paper describes the use
		  of the Hamming learning vector quantizer (H-LVQ), which is
		  used to generate a codebook of speech vectors in which such
		  a correlation exists.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cawley96a,
  author	= {G. C. Cawley},
  title		= {An improved vector quantisation algorithm for speech
		  transmission over noisy channels},
  booktitle	= {Proceedings ICSLP 96. Fourth International Conference on
		  Spoken Language Processing},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {1},
  editor	= {H. T. Bunnell and W. Idsardi},
  address	= {New York, NY, USA},
  pages		= {299--301},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cazuguel98a,
  author	= {G. Cazuguel and A. Cziho and B. Solaiman and C. Roux},
  title		= {Medical image compression and characterization using
		  vector quantization: an application of
		  \mbox{self-organizing} maps and quadtree decomposition},
  booktitle	= {Proceedings. 1998 IEEE International Conference on
		  Information Technology Applications in Biomedicine, ITAB
		  '98},
  publisher	= {IEEE},
  year		= {1998},
  editor	= {S. Laxminarayan and E. Micheli-Tzanakou},
  address	= {New York, NY, USA},
  pages		= {127--32},
  dbinsdate	= {oldtimer}
}

@Article{	  cereghino01a,
  author	= {Cereghino, R. and Giraudel, J. L. and Compin, A.},
  title		= {Spatial analysis of stream invertebrates distribution in
		  the Adour-Garonne drainage basin (France), using Kohonen
		  self organizing maps},
  journal	= {ECOLOGICAL MODELLING},
  year		= {2001},
  volume	= {146},
  number	= {1--3},
  month		= {DEC 1},
  pages		= {167--180},
  abstract	= {We analysed the regional distribution of 283 lotic
		  macroinvertebrate species from four insect orders
		  (Ephemeroptera, Plecoptera, Trichoptera, Coleoptera = EPTC)
		  in the Adour-Garonne drainage basin (South-Western France,
		  surface = 116 000 km(2)). The aim of this work was to
		  provide a stream classification based on characteristic
		  species assemblages. The faunistic data corresponded to the
		  occurrence (presence or absence) of 283 species at 252
		  sampling sites. These data were computed with the Kohonen
		  self organised map algorithm (SOM) (Kohonen,
		  Self-Organizing Maps, volume30 of Springer Series in
		  Information Sciences. Springer, Berlin, Heidelberg. (Second
		  Extended Edition 1997)). This neural network algorithm has
		  already been successfully used in ecology (Giraudel et al.,
		  Artificial neural networks, applications to ecology and
		  evolution. Springer-Verlag, (in press); Chon et al., Ecol.
		  Model., 90, 1996, 69--78) for communities patternizing. SOM
		  enable visualisation of the complex species assemblage in a
		  two-dimensional space, preserving the topology of the input
		  data. Then, using the U-matrix method, it was possible to
		  classify the data without prior knowledge. Four major EPTC
		  regions were characterised within the drainage basin
		  (Massif Central mountains, Pyrenees mountains, Piedmont and
		  plains, Toulouse city agglomeration), along with their
		  theoretical species assemblage. The number of species
		  characterising each region ranged from 45 to 159,
		  underlining the spatial (i.e. longitudinal and
		  geographical) differences in EPTC assemblages. The main
		  interest of our results is that the stability of these
		  theoretical assemblages may be used to define
		  representative and/or reference sites for biological
		  surveillance, as any change in species composition within a
		  given EPTC region can be considered as a biological
		  indicator of environmental changes. },
  dbinsdate	= {2002/1}
}

@InProceedings{	  cervera95a,
  author	= {Enrique Cervera and Angel P. {del Pobil}},
  title		= {Perception-based Qualitative Reasoning in Manipulation
		  with Uncertainty},
  booktitle	= {Proc. CAEPIA'95, VI Conference of the Spanish Association
		  for Artificial Intelligence},
  year		= {1995},
  pages		= {129--139},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cervera95b,
  author	= {Enrique Cervera and Angel P. {del Pobil} and Edward Marta
		  and Miguel A. Serna},
  title		= {Interpreting Tactile Information with Neural Networks in
		  Robot Tasks},
  booktitle	= {Proc. CAEPIA'95, VI Conference of the Spanish Association
		  for Artificial Intelligence},
  year		= {1995},
  pages		= {415--423},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cervera95c,
  author	= {Enrique Cervera and Angel P. {del Pobil}},
  title		= {A Supervised Learning Method with Multiple Self-Organizing
		  Maps},
  booktitle	= {Proc. CAEPIA'95, VI Conference of the Spanish Association
		  for Artificial Intelligence},
  year		= {1995},
  pages		= {471--479},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cervera95d,
  author	= {Enrique Cervera and Angel P. {del Pobil} and Edward Marta
		  and Miguel A. Serna},
  title		= {Monitoring Robotic Tasks in a Floxible Manufacturing
		  System},
  booktitle	= {Proc. TTIA'95, Transferencia Tecnol{\'{o}}gica de
		  Inteligencia Artificial a Industria, Medicina y
		  Aplicaciones Sociales},
  year		= {1995},
  editor	= {Ram{\'{o}}n Rizo Aldeguer and Juan Manuel Gar{\'{c}}ia
		  Chamizo},
  pages		= {3--12},
  dbinsdate	= {oldtimer}
}

@Article{	  cervera95e,
  author	= {Cervera, E. and {del Pobil}, A. P. },
  title		= {On the integration of sensors and neural networks in
		  intelligent robotic systems},
  journal	= {Systems Analysis Modelling Simulation},
  year		= {1995},
  volume	= {18--19},
  pages		= {297--300},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@Article{	  cervera95f,
  author	= {Cervera, E. and {del Pobil}, A. P. and Marta, E. and
		  Serna, M. A. },
  title		= {Use of sensors to deal with uncertainty in realistic
		  robotic environments},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1995},
  volume	= {2492},
  number	= {pt. 2},
  pages		= {740--7},
  publisher	= {SPIE-Int. Soc. Opt. Eng},
  annote	= {A conference paper in journal},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cervera95g,
  author	= {E. Cervera and A. P. {del Pobil}},
  title		= {Multiple \mbox{self-organizing} maps for supervised
		  learning},
  booktitle	= {From Natural to Artificial Neural Computation.
		  International Workshop on Artificial Neural Networks.
		  Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1995},
  editor	= {J. Mira and F. Sandoval},
  address	= {Berlin, Germany},
  pages		= {345--52},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cervera95h,
  author	= {E. Cervera and A. P. {del Pobil} and E. Marta and M. A.
		  Serna},
  title		= {Dealing with uncertainty in fine motion: a neural
		  approach},
  booktitle	= {Industrial and Engineering Applications of Artificial
		  Intelligence and Expert Systems. Proceedings of the Eighth
		  International Conference},
  publisher	= {Gordon \& Breach},
  year		= {1995},
  editor	= {G. F. Forsyth and M. Ali},
  address	= {Newark, NJ, USA},
  pages		= {119--26},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cervera95i,
  author	= {E. Cervera and A. P. {del Pobil}},
  title		= {Self-organizing maps for supervision in robot pick-and-
		  place operations},
  booktitle	= {Artificial Neural Nets and Genetic Algorithms. Proceedings
		  of the International Conference},
  publisher	= {Springer-Verlag},
  year		= {1995},
  editor	= {D. W. Pearson and N. C. Steele and R. F. Albrecht},
  address	= {Vienna, Austria},
  pages		= {372--5},
  dbinsdate	= {oldtimer}
}

@Article{	  cervera96a,
  author	= {E. Cervera and A. P. {del Pobil} and E. Marta and M. A.
		  Serna},
  title		= {Perception-based learning for motion in contact in task
		  planning},
  journal	= {Journal of Intelligent and Robotic Systems: Theory and
		  Applications},
  year		= {1996},
  volume	= {17},
  number	= {3},
  pages		= {283--308},
  dbinsdate	= {oldtimer}
}

@Article{	  cervera97a,
  author	= {E. Cervera and A. P. {del Pobil}},
  title		= {Multiple \mbox{self-organizing} maps: a hybrid learning
		  scheme},
  journal	= {Neurocomputing},
  year		= {1997},
  volume	= {16},
  number	= {4},
  pages		= {309--18},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cervera99a,
  author	= {E. Cervera and A. P. {del Pobil}},
  title		= {A SOM-based sensing approach to robotic manipulation
		  tasks},
  booktitle	= {Kohonen Maps},
  pages		= {207--218},
  publisher	= {Elsevier},
  year		= {1999},
  editor	= {Oja, E. and Kaski, S.},
  address	= {Amsterdam},
  annote	= {Keywords: robots, manipulation, force sensing, learning},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cetin95a,
  author	= {D. {\c{C}}etin and F. Yildirim and D. Demirekler and B.
		  Nakibo{\u{g}}lu and B. T{\"{u}}z{\"{u}}n},
  title		= {Text-Independent Speaker Identification Using Learning
		  Vector Quantization},
  booktitle	= {Proc. EANN'95, Engineering Applications of Artificial
		  Neural Networks},
  year		= {1995},
  pages		= {267--269},
  organization	= {Finnish Artificial Intelligence Society},
  dbinsdate	= {oldtimer}
}

@Article{	  cha01a,
  author	= {Cha, Y. Y. and Oh, J. H.},
  title		= {A field-of-view generation algorithm using neural
		  network},
  journal	= {Mechatronics},
  year		= {2001},
  volume	= {11},
  number	= {6},
  month		= {September },
  pages		= {731--744},
  organization	= {Dept. of Mech. Design Engineering, College of Engineering,
		  Won-Kwang University},
  publisher	= {},
  address	= {},
  abstract	= {In various works using vision system, if a field of view
		  (FOV) of CCD camera cannot cover overall working area on a
		  table or a floor, the overall working area must be divided
		  into several areas whose sizes equal that of FOV. In this
		  study, an effective FOV generation algorithm in which FOV
		  is automatically made is proposed for vision system.
		  Instead of existing sequential FOV generation method, we
		  propose a new FOV generation method by using newly modified
		  self-organizing map (SOM) which has multiple structure
		  based on SOM of neural network and uses new training rule
		  by introducing the movement, creation and deletion terms.
		  In consequence of real experiment, we show that the
		  proposed algorithm is able to decrease the number of
		  created FOVs about 10% more than the existing sequential
		  algorithm. This result is acquired by which the proposed
		  algorithm minimizes overlaps between FOVs, assigns optimal
		  positions of FOVs and eliminates the unnecessary FOVs.
		  Consequently, as the number of FOVs decreases, we have been
		  able to diminish the total working time using vision
		  system.},
  dbinsdate	= {2002/1}
}

@Article{	  cha97a,
  author	= {Jihun Cha and L. V. Fausett},
  title		= {Comparison of three clustering algorithms and an
		  application to color image compression},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1997},
  volume	= {3077},
  pages		= {225--35},
  note		= {(Applications and Science of Artificial Neural Networks
		  III Conf. Date: 21--24 April 1997 Conf. Loc: Orlando, FL,
		  USA Conf. Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cha98a,
  author	= {Eui Young Cha and Myung Ho Kang},
  title		= {Multiple target tracking in clutter backgrounds using
		  \mbox{self-organizing} feature map},
  booktitle	= {1998 IEEE International Joint Conference on Neural
		  Networks Proceedings. IEEE World Congress on Computational
		  Intelligence},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {2},
  address	= {New York, NY, USA},
  pages		= {1162--6},
  dbinsdate	= {oldtimer}
}

@Article{	  chakraborty00a,
  author	= {Chakraborty, G. and Chakraborty, B.},
  title		= {A novel normalization technique for unsupervised learning
		  in {ANN}},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {2000},
  volume	= {11},
  pages		= {253--7},
  abstract	= {Unsupervised learning is used to categorize
		  multidimensional data into a number of meaningful classes
		  on the basis of the similarity or correlation between
		  individual samples. In neural-network implementation of
		  various unsupervised algorithms such as principal component
		  analysis, competitive learning or self-organizing map,
		  sample vectors are normalized to equal lengths so that
		  similarity could be easily and efficiently obtained by
		  their dot products. In general, sample vectors span the
		  whole multidimensional feature space and existing
		  normalization methods distort the intrinsic patterns
		  present in the sample set. In this work, a novel method of
		  normalization by mapping the samples to a new space of one
		  more dimension is proposed. The original distribution of
		  the samples in the feature space is shown to be almost
		  preserved in the transformed space. Simple rules are given
		  to map from original space to the normalized space and vice
		  versa.},
  dbinsdate	= {oldtimer}
}

@Article{	  chakraborty93a,
  author	= {K. Chakraborty and U. Roy},
  title		= {Connectionist models for part-family classifications},
  journal	= {Computers \& Industrial Engineering},
  year		= {1993},
  volume	= {24},
  number	= {2},
  pages		= {189--198},
  month		= {April},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan00a,
  author	= {Chan, A. and Spracklen, T.},
  title		= {Feature indicators: a self-organizing map approach to
		  legacy code},
  booktitle	= {Proceedings of the International Conference on Artificial
		  Intelligence. IC-AI'2000. CSREA Press, Athens, GA, USA},
  year		= {2000},
  volume	= {3},
  pages		= {1449--54},
  abstract	= {The self-organizing map's unsupervised clustering
		  property, is known for classifying high dimensional data
		  sets into clusters that have similar features. Based on
		  this property, it is demonstrated that a self-organizing
		  map is capable of identifying features within software code
		  by grouping procedures with similar properties together.
		  This permits us to identify potential objects, abstract
		  data types or classes. To further enhance the visualization
		  of clusters by providing a form of justification for the
		  groupings, we introduce "feature indicators" that labels
		  the map with the elements that most contributed to each
		  fired self-organizing map's unit.},
  dbinsdate	= {2002/1}
}

@Article{	  chan01a,
  author	= {Chan, C. W. and Jin, H. and Cheung, K. C. and Zhang, H.
		  Y.},
  title		= {Fault detection of systems with redundant sensors using
		  constrained Kohonen networks},
  journal	= {Automatica},
  year		= {2001},
  volume	= {37},
  number	= {10},
  month		= {October },
  pages		= {1671--1676},
  organization	= {Department of Mechanical Engineering, University of Hong
		  Kong},
  publisher	= {},
  address	= {},
  abstract	= {The Kohonen self-organizing map (KN) was developed for
		  pattern recognition, and has been extended to fault
		  classification. However, the KN cannot be applied to
		  classify faults from the system output if it contains other
		  factors, such as system state and sensor mounting errors.
		  To overcome this problem, a constrained KN (CKN) is
		  proposed. To eliminate the effect of the system state and
		  the mounting errors, it is proposed that the weight vectors
		  of the CKN are constrained in the parity space. The
		  training algorithm of the CKN is derived, and its
		  convergence discussed. Application of the CKN to fault
		  classification is presented, and its performance is
		  illustrated by an example involving a redundant sensor
		  system with six sensors.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chan01b,
  author	= {Chan, J. C. W. and DeFries R. S. and Townshend, J. R. G.},
  title		= {Improved recognition of spectrally mixed land cover
		  classes using spatial textures and voting classifications},
  booktitle	= {Computer Analysis of Images and Patterns. 9th
		  International Conference, CAIP 2001. Proceedings (Lecture
		  Notes in Computer Science Vol.2124). Springer-Verlag,
		  Berlin, Germany},
  year		= {2001},
  volume	= {},
  pages		= {217--27},
  abstract	= {Forest regeneration is important to account for carbon
		  sinks. Mapping regenerating forests from satellite data is
		  difficult because they are spectrally mixed with natural
		  forests. This paper investigates the combined use of
		  texture features and voting classifications to enhance the
		  recognition of these two classes. Bagging and boosting were
		  applied to learning vector quantization (LVQ) and decision
		  trees. Our results showed that spatial textures improved
		  the class separability. After applying voting
		  classifications, the class accuracy of the decision tree
		  increased by 5--7% and that of LVQ by approximately 3%. A
		  substantial reduction (between 23% to 40%) of confusions
		  between regenerating forests and natural forests was
		  recorded. Comparatively, bagging is more consistent than
		  boosting. An interesting observation is that even LVQ, a
		  stable learner, was able to benefit from both voting
		  classification algorithms.},
  dbinsdate	= {2002/1}
}

@Article{	  chan01d,
  author	= {Chan, J. C. W. and Chan, K. P. and Yeh, A. G. O.},
  title		= {Detecting the nature of change in an urban environment: A
		  comparison of machine learning algorithms},
  journal	= {PHOTOGRAMMETRIC ENGINEERING AND REMOTE SENSING},
  year		= {2001},
  volume	= {67},
  number	= {2},
  month		= {FEB},
  pages		= {213--225},
  abstract	= {The performance of difference machine learning algorithms
		  for detecting nature of change was compared. To alleviate
		  the problem of obtaining enough training data, simulated
		  training data were generated from single-date images. A
		  one-pass classification with four machine learning
		  algorithms, namely, Multi-Layer Perceptrons (MLP), Learning
		  Vector Quantization (LVQ), Decision Tree Classifiers (DTC),
		  and the Maximum- Likelihood Classifier (MLC), were tested.
		  Recognition rates, ease of use, and degree of automation of
		  the four algorithms were assessed. The results showed that
		  the incorporation of cross-combined simulated training data
		  enhanced the detection of nature of change. Compared to
		  conventional post- classification comparison methods, LVQ
		  and ore did better in terms of overall accuracy In terms of
		  average accuracy of the change classes, LVQ was the best
		  performer. DTC was the easiest to use and the most robust
		  in training. MLP procedures were the most difficult to
		  replicate.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chan01e,
  author	= {Chan, A. T. S. and Shiu, A. and Jiannong Cao and Hong Va
		  Leong},
  title		= {Reactive web policing based on self-organizing maps},
  booktitle	= {Proceedings of IEEE Region 10 International Conference on
		  Electrical and Electronic Technology. TENCON 2001. IEEE,
		  Piscataway, NJ, USA},
  year		= {2001},
  volume	= {1},
  pages		= {160--4},
  abstract	= {Almost without any doubt, the Internet and Web in
		  particular, have brought about radical changes in
		  information retrieval with unparallel benefits to
		  organizations and individuals to gain access to vast amount
		  of articles and documents. The focus on restricting
		  inappropriate materials at their source is not well suited
		  to the nature and open architecture of Internet, where
		  information source maybe in a different legal jurisdiction
		  than the recipient. This paper proposes a reactive approach
		  based on SOM (Self-Organising Maps) neural network to deal
		  with dynamic blocking and filtering of Internet contents.
		  We describe the design and implementation of a web policing
		  proxy (WebPolice) based on a Kohonen's neural network model
		  that attempts to classify the web contents dynamically
		  using competitive learning. The parameter setting of the
		  network has been experimented to obtain the optimal
		  classification rate and performance for the model.},
  dbinsdate	= {2002/1}
}

@Article{	  chan93a,
  author	= {Chan, L. S. -C. and Hean-Lee Poh and Teo Jasic},
  title		= {Neural networks and their applications},
  journal	= {Computer Processing of Chinese \& Oriental Languages},
  year		= {1993},
  volume	= {7},
  number	= {2},
  pages		= {133--66},
  month		= {Dec},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan94a,
  author	= {Mike V. Chan and Xin Feng and James A. Heinen and Russell
		  J. Niederjohn},
  title		= {Classification of Speech Accents with Neural Networks},
  pages		= {4483--4486},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, speech recognition, comparison},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chan95a,
  author	= {L. A. Chan and N. M. Nasrabadi and V. Mirelli},
  title		= {Automatic target recognition using modularly cascaded
		  vector quantizers and multilayer perceptrons},
  booktitle	= {1996 IEEE International Conference on Acoustics, Speech,
		  and Signal Processing Conference Proceedings},
  publisher	= {ASME Press},
  year		= {1995},
  volume	= {6},
  editor	= {C. H. Dagli and M. Akay and C. L. P. Chen and B. R.
		  Fernandez and J. Ghosh},
  address	= {New York, NY, USA},
  pages		= {3386--9},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan95b,
  author	= {Samuel W. K. Chan and James Franklin},
  title		= {A Neurosymbolic Integrated Model for Semantic Ambiguation
		  Resolution},
  volume	= {VI},
  pages		= {2965--2970},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan95c,
  author	= {K. W. Chan and K. L. Chan},
  title		= {Multi-Reference Neighborhood Search for Vector
		  Quantization by Neural Network Prediction and
		  Self-Organizing Feature Map},
  volume	= {IV},
  pages		= {1898--1902},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan95d,
  author	= {Chan, K. W. and Chan, K. L. },
  title		= {Multi-reference neighborhood search for vector
		  quantization by self-organized featured map},
  booktitle	= {Fifth International Conference on Image Processing and its
		  Applications},
  year		= {1995},
  pages		= {579--83},
  organization	= {City Polytech. of Hong Kong, Kowloon, Hong Kong},
  publisher	= {IEE},
  address	= {London, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan95e,
  author	= {Lai-Wan Chan and Man-Wai Chau and Wing-Chung Chung},
  title		= {GlobalOR: a parallel implementation of the
		  \mbox{self-organizing} map},
  booktitle	= {Proceedings of International Conference on Neural
		  Information Processing (ICONIP `95)},
  year		= {1995},
  editor	= {Zhong, Y. and Yang, Y. and Wang, M. },
  volume	= {2},
  pages		= {625--8},
  organization	= {Dept. of Comput. Sci. , Chinese Univ. of Hong Kong,
		  Shatin, Hong Kong},
  publisher	= {Publishing House of Electron. Ind},
  address	= {Beijing, China},
  dbinsdate	= {oldtimer}
}

@Article{	  chan96a,
  author	= {L. A. Chan and N. H. Nasrabadi and V. Mirelli},
  title		= {Wavelet-based learning vector quantization for automatic
		  target recognition},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1996},
  volume	= {2755},
  pages		= {82--93},
  note		= {(Signal Processing, Sensor Fusion, and Target Recognition
		  V Conf. Date: 8--10 April 1996 Conf. Loc: Orlando, FL, USA
		  Conf. Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chan96b,
  author	= {L. A. Chan and N. M. Nasrabadi and V. Mirelli},
  title		= {Multi-stage target recognition using modular vector
		  quantizers and multilayer perceptrons},
  booktitle	= {Proceedings 1996 IEEE Computer Society Conference on
		  Computer Vision and Pattern Recognition},
  publisher	= {IEEE Computer Society Press},
  year		= {1996},
  address	= {Los Alamitos, CA, USA},
  pages		= {114--19},
  abstract	= {An automatic target recognition (ATR) classifier is
		  proposed that uses modularly cascaded vector quantizers
		  (VQs) and multilayer perceptrons (MLPs). A dedicated VQ
		  codebook is constructed for each target class at a specific
		  range of aspects, which is trained with the K-means
		  algorithm and a modified learning vector quantization (LVQ)
		  algorithm. Each final codebook is expected to give the
		  lowest mean squared error (MSE) for its correct target
		  class at a given range of aspects. These MSEs are then
		  processed by an array of window MLPs and a target MLP
		  consecutively. In the spatial domain target recognition
		  rates of 90.3 and 65.3 percent are achieved for moderately
		  and highly cluttered test sets, respectively. Using the
		  wavelet decomposition with an adaptive and independent
		  codebook per subband, the VQs alone have produced
		  recognition rates of 98.7 and 69.0 percent on more
		  challenging training and test sets, respectively.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chan96c,
  author	= {L. A. Chan and N. M. Nasrabadi},
  title		= {Modular wavelet-based vector quantization for automatic
		  target recognition},
  booktitle	= {1996 IEEE/SICE/RSJ International Conference on Multisensor
		  Fusion and Integration for Intelligent Systems},
  publisher	= {IEEE},
  year		= {1996},
  address	= {New York, NY, USA},
  pages		= {462--9},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chan96d,
  author	= {L. A. Chan and N. M. Nasrabadi},
  title		= {An application of wavelet-based vector quantization in
		  target recognition},
  booktitle	= {Proceedings of the IEEE International Joint Symposia on
		  Intelligence and Systems},
  publisher	= {IEEE Computer Society Press},
  year		= {1996},
  address	= {Los Alamitos, CA, USA},
  pages		= {274--81},
  abstract	= {An automatic target recognition (ATR) classifier is
		  constructed that uses a set of dedicated vector quantizers
		  (VQs). The background pixels in each input image are
		  properly clipped out by a set of aspect windows. The
		  extracted target area for each aspect window is then
		  enlarged to a fixed size, after which a wavelet
		  decomposition splits the enlarged extraction into several
		  subbands. A dedicated VQ codebook is generated for each
		  subband of a particular target class at a specific range of
		  aspects. Thus, each codebook consists of a set of feature
		  templates that are iteratively adapted to represent a
		  particular subband of a given target class at a specific
		  range of aspects. These templates are then further trained
		  by a modified learning vector quantization (LVQ) algorithm
		  that enhances their discriminatory characteristics.},
  dbinsdate	= {oldtimer}
}

@Article{	  chan97a,
  author	= {L. A. Chan and N. M. Nasrabadi},
  title		= {An application of wavelet-based vector quantization in
		  target recognition},
  journal	= {International Journal on Artificial Intelligence Tools
		  [Architectures, Languages, Algorithms]},
  year		= {1997},
  volume	= {6},
  number	= {2},
  pages		= {165--78},
  dbinsdate	= {oldtimer}
}

@Book{		  chan97b,
  author	= {Chan, L. A. and Nasrabadi, N. M.},
  title		= {Automatic Target Recognition Using Wavelet-Based Vector
		  Quantization. Interim rept. Dec 96-Jul 97.},
  year		= {1997},
  abstract	= {An automatic target recognition classifier is described
		  that uses a set of dedicated vector quantizers (VQs) in the
		  wavelet domain. The background pixels in each input image
		  are properly clipped out by a set of aspect windows. The
		  extracted target area for each aspect window is then
		  enlarged to a fixed size, after which a wavelet
		  decomposition is used to split this region into several
		  subbands. A dedicated VQ codebook is then generated for
		  each subband of a particular target class at a specific
		  range of aspects. Thus, each codebook consists of a set of
		  feature templates that are iteratively adapted to represent
		  a particular subband of a given target class at a specific
		  range of aspects. These templates are then further trained
		  by a modified learning vector quantization (LVQ) algorithm
		  that enhances their discriminatory characteristics.
		  Finally, a path selector was designed to speed up the
		  recognition process at the expense of a tolerable
		  degradation in the recognition rate.},
  dbinsdate	= {oldtimer}
}

@Article{	  chan99a,
  author	= {Chan, L. A. and Nasrabadi, N. M.},
  title		= {Automatic target recognition using vector quantization and
		  neural networks},
  journal	= {Optical-Engineering},
  year		= {1999},
  volume	= {38},
  pages		= {2147--61},
  abstract	= {We propose an automatic target recognition (ATR) algorithm
		  that uses a set of dedicated vector quantizers (VQs) and
		  multilayer perceptrons (MLPs). For each target class at a
		  specific range of aspects, the background pixels of an
		  input image are first removed. The extracted target area is
		  then subdivided into several subimages. A dedicated VQ
		  codebook is constructed for each of the resulting
		  subimages. Using the K-means algorithm, each VQ codebook
		  learns a set of patterns representing the local features of
		  a particular target for a specific range of aspects. The
		  resulting codebooks are further trained by a modified
		  learning vector quantization (LVQ) algorithm, which
		  enhances the discriminatory power of the codebooks. Each
		  final codebook is expected to give the lowest mean squared
		  error (MSE) for its correct target class and range of
		  aspects. These MSEs are then input to an array of
		  window-level MLPs (WMLPs), where each WMLP is specialized
		  in recognizing its intended target class for a specific
		  range of aspects. The outputs of these WMLPs are
		  manipulated and passed to a target-level MLP, which
		  produces the final recognition results. We trained and
		  tested the proposed ATR algorithm on large and realistic
		  data sets and obtained impressive results using the
		  wavelet-based adaptive product VQs configuration.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chan99b,
  author	= {Chan, C. W. and Cheung, K. C. and Hong Jin and Zhang, H.
		  Y.},
  title		= {A constrained Kohonen network and its application to
		  sensor fault detection},
  booktitle	= {Automatic Control in Aerospace 1998. Proceedings volume
		  from the 14th IFAC Symposium. Elsevier Science, Kidlington,
		  UK},
  year		= {1999},
  volume	= {},
  pages		= {249--54},
  abstract	= {Kohonen network (KN) has a good capability for pattern
		  classification. However when measurements used for this
		  purpose consist not only of the pattern to be classified
		  but also of other information, the original KN can not be
		  used successfully. For example, the sensor measurement data
		  not only consist of fault information we want to classify
		  but also information of system state variable and other
		  errors and noises. In order to detect and isolate faults in
		  the sensor measurement a constrained Kohonen network (CKN)
		  is constructed and its algorithm is developed in this
		  paper. The main idea of CKY is to constrain the weight
		  vector of the network in some subspace such that the
		  unwanted information (e.g., that of the system state
		  variable) is zeroed by the weight vector while the wanted
		  information (e.g., faults) is classified. The CKN can be
		  successfully used to detect and isolate faults using sensor
		  measurements.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chance01a,
  author	= {Chance, D. and Lebby, G. L. and King, M. C. and Cooke, M.
		  J.},
  title		= {Unsupervised mapping of biometric signatures for enhanced
		  computer security},
  booktitle	= {Proceedings of the IASTED International Conference.
		  Artificial Intelligence and Applications. ACTA Press,
		  Anaheim, CA, USA},
  year		= {2001},
  volume	= {},
  pages		= {86--8},
  abstract	= {Describes a method of performing biometric security to
		  enhance the ability to authenticate a user given a known
		  reliable biometric database. We propose a clinical
		  technique called nerve conduction and apply artificial
		  neural networking technology in the form of the Kohonen
		  paradigm to enhance automated biometric security. The nerve
		  conduction study is applied to measure the nerve response
		  to electrical stimulation across a given pathway. By
		  applying this procedure we obtain a biological signal of a
		  user, which is interpreted as a biometric print of user
		  uniqueness. The use of the Kohonen method assures that the
		  high level of distinct classification desired is achieved.
		  Experimental results imply that this approach is highly
		  effective and the implementation of real-time data
		  extraction and analysis lend this method to application in
		  continuous high traffic secure environments.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chandrasekaran93a,
  author	= {V. Chandrasekaran and M. Palaniswami and Terry M. Caelli},
  title		= {Performance Evaluation of Spatio-Temporal Feature Maps
		  with gated Neuronal Architecture},
  booktitle	= {Proc. WCNN'93, World Congress on Neural Networks},
  year		= {1993},
  volume	= {IV},
  pages		= {112--118},
  organization	= {{INNS}},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chandrasekaran93b,
  author	= {V. Chandrasekaran and M. Palaniswami and Terry M. Caelli},
  title		= {An Extended Self-Organizing Map with Gated Neurons},
  booktitle	= {Proc. ICNN'93, International Conference on Neural
		  Networks},
  year		= {1993},
  volume	= {III},
  pages		= {1474--1479},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chandrasekaran95a,
  author	= {Chandrasekaran, V. and Palaniswami, M. and Caelli, T. M.
		  },
  title		= {Pattern recognition by topology free spatio-temporal
		  feature map},
  booktitle	= {1995 IEEE International Conference on Systems, Man and
		  Cybernetics. Intelligent Systems for the 21st Century},
  year		= {1995},
  volume	= {2},
  pages		= {1136--41},
  organization	= {Sch. of Electr. Eng. \& Comput. Sci. , Melbourne Univ. ,
		  Parkville, Vic. , Australia},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  chandrasekaran95b,
  author	= {Chandrasekaran, V. and Palaniswami, M. and Caelli, T. M.},
  title		= {Spatio-temporal feature maps using gated neuronal
		  architecture},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1995},
  volume	= {6},
  number	= {5},
  pages		= {1119--31},
  month		= {Sept},
  abstract	= {In this paper, Kohonen's self-organizing feature map is
		  modified by a novel technique of allowing the neurons in
		  the feature map to compete in a selective manner. The
		  selective competition is achieved by grating the
		  N-dimensional feature space using a spatial frequency and
		  setting a criterion for the neurons to compete based on the
		  region in which the input pattern resides. The spatial
		  grating and selective competition are achieved by
		  introducing a gated neuronal architecture in the feature
		  map. As the selection criterion changes with time, it
		  generates a time sequence of winning node indexes providing
		  more input information and potentially allowing higher
		  classification performance. These time sequences are then
		  used to predict the class label of the input pattern more
		  accurately. Three possible class label prediction
		  algorithms are formulated based on evidential reasoning
		  method and Bayes conditional probability theorem. These are
		  tested on real world 8-class texture and a synthetic
		  12-class three-dimensional object recognition problems. The
		  classification performance is then compared with the
		  results obtained by using a standard statistical linear
		  discriminant analysis.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chandrasekaran97a,
  author	= {V. Chandrasekaran and Zhi-Qiang Liu},
  title		= {Projection pursuits in {SOM} classifiers},
  booktitle	= {Proceedings of WSOM'97, Workshop on Self-Organizing Maps,
		  Espoo, Finland, June 4--6},
  publisher	= {Helsinki University of Technology, Neural Networks
		  Research Centre},
  year		= 1997,
  address	= {Espoo, Finland},
  pages		= {100--105},
  dbinsdate	= {oldtimer}
}

@Article{	  chandrasekaran98a,
  author	= {Chandrasekaran, V. and Liu, Zhi Qiang},
  title		= {Topology constraint free fuzzy gated neural networks for
		  pattern recognition},
  journal	= {IEEE Transactions on Neural Networks},
  year		= {1998},
  number	= {3},
  volume	= {9},
  pages		= {483--502},
  abstract	= {In this paper, a novel topology constraint free neural
		  network architecture using a generalized fuzzy gated neuron
		  model is presented for pattern recognition task. The main
		  feature is that the network does not require weight
		  adaptation at its input and the weights are initialized
		  directly from the training pattern set. The elimination of
		  the need for iterative weight adaptation schemes
		  facilitates quick network set up times which make the fuzzy
		  gated neural networks very attractive. The performance of
		  the proposed network is found to be functionally equivalent
		  to spatio-temporal feature maps under a mild technical
		  condition. The classification performance of fuzzy gated
		  neural network is demonstrated on a 12-class synthetic
		  three-dimensional (3-D) object data set, real-world
		  eight-class texture data set, and real-world 12-class 3-D
		  object data set. The performance results are compared with
		  the classification accuracies obtained from spatio-temporal
		  feature map, adaptive subspace self-organizing map,
		  multilayer feedforward neural networks, radial basis
		  function neural networks, and linear discriminant analysis.
		  Despite the network's ability to accurately classify seen
		  data and adequately generalize validation data, its
		  performance is found to be sensitive to noise perturbations
		  due to fine fragmentation of the feature space. This paper
		  also provides partial solutions to the above robustness
		  issue by proposing certain improvements to various modules
		  of the proposed fuzzy gated neural network.},
  dbinsdate	= {oldtimer}
}

@Article{	  chang00a,
  author	= {Chang, Jyh-Shan and Chiueh, Tzi-Dar},
  title		= {Image vector quantization using classified
		  binary-tree-structured self-organizing feature maps},
  journal	= {IEICE Transactions on Information and Systems},
  year		= {2000},
  volume	= {E-83-D},
  number	= {10},
  month		= {Oct},
  pages		= {1898--1907},
  organization	= {Natl Taiwan Univ},
  publisher	= {IEICE of Japan},
  address	= {Tokyo},
  abstract	= {With the continuing growth of the World Wide Web (WWW)
		  services over the Internet, the demands for rapid image
		  transmission over a network link of limited bandwidth and
		  economical image storage of a large image database are
		  increasing rapidly. In this paper, a classified
		  binary-tree-structured Self Organizing Feature Map neural
		  network is proposed to design image vector codebooks for
		  quantizing images. Simulations show that the algorithm not
		  only produces codebooks with lower distortion than the
		  well-known CVQ algorithm but also can minimize the edge
		  degradation. Because the adjacent codewords in the proposed
		  algorithm are updated concurrently, the codewords in the
		  obtained codebooks tend to be ordered according to their
		  mutual similarity which means more compression can be
		  achieved with this algorithm. It should also be noticed
		  that the obtained codebook is particularly well suited for
		  progressive image transmission because it always forms a
		  binary tree in the input space.},
  dbinsdate	= {2002/1}
}

@Article{	  chang01a,
  author	= {Chang, Kui-yu and Ghosh, Joydeep},
  title		= {Unified model for probabilistic principal surfaces},
  journal	= {IEEE Transactions on Pattern Analysis and Machine
		  Intelligence},
  year		= {2001},
  volume	= {23},
  number	= {1},
  month		= {Jan},
  pages		= {22--41},
  organization	= {Interwoven, Inc},
  publisher	= {IEEE},
  address	= {Los Alamitos, CA},
  abstract	= {Principal curves and surfaces are nonlinear
		  generalizations of principal components and subspaces,
		  respectively. They can provide insightful summary of
		  high-dimensional data not typically attainable by classical
		  linear methods. Solutions to several problems, such as
		  proof of existence and convergence, faced by the original
		  principal curve formulation have been proposed in the past
		  few years. Nevertheless, these solutions are not generally
		  extensible to principal surfaces, the mere computation of
		  which presents a formidable obstacle. Consequently,
		  relatively few studies of principal surfaces are available.
		  Recently, we proposed the probabilistic principal surface
		  (PPS) to address a number of issues associated with current
		  principal surface algorithms. PPS uses a manifold oriented
		  covariance noise model, based on the generative
		  topographical mapping (GTM), which can be viewed as a
		  parametric formulation of Kohonen's self-organizing map.
		  Building on the PPS, we introduce a unified covariance
		  model that implements PPS (0 < \alpha < 1), GTM (\alpha =
		  1), and the manifold-aligned GTM (\alpha > 1) by varying
		  the clamping parameter \alpha. Then, we comprehensively
		  evaluate the empirical performance (reconstruction error)
		  of PPS, GTM, and the manifold-aligned GTM on three popular
		  benchmark data sets. It is shown in two different
		  comparisons that the PPS outperforms the GTM under
		  identical parameter settings. Convergence of the PPS is
		  found to be identical to that of the GTM and the
		  computational overhead incurred by the PPS decreases to 40
		  percent or less for more complex manifolds. These results
		  show that the generalized PPS provides a flexible and
		  effective way of obtaining principal surfaces.},
  dbinsdate	= {2002/1}
}

@Article{	  chang02a,
  author	= {Chang, H. C. and Kopaska-Merkel, D. C. and Chen, H. C.},
  title		= {Identification of lithofacies using Kohonen
		  self-organizing maps},
  journal	= {COMPUTERS \& GEOSCIENCES},
  year		= {2002},
  volume	= {28},
  number	= {2},
  month		= {MAR},
  pages		= {223--229},
  abstract	= {Lithofacies identification is a primary task in reservoir
		  characterization. Traditional techniques of lithofacies
		  identification from core data are costly, and it is
		  difficult to extrapolate to non-cored wells. We present a
		  low-cost automated technique using Kohonen self-organizing
		  maps (SOMs) to identify systematically and objectively
		  lithofacies from well log data, SOMs are unsupervised
		  artificial neural networks that map the input space into
		  clusters in a topological form whose organization is
		  related to trends in the input data. A case study used five
		  wells located in Appleton Field. Escambia County, Alabama
		  (Smackover Formation, limestone and dolomite. Oxfordian,
		  Jurassic). A five-input. one-dimensional output approach is
		  employed, assuming the lithofacies are in
		  ascending/descending order with respect to
		  paleoenvironmental energy levels. To consider the possible
		  appearance of new logfacies not seen in training mode,
		  which may potentially appear in test wells, the maximum
		  number of outputs is set to 20 instead of four, the
		  designated number of lithofacies in the study area. This
		  study found eleven major clusters. The clusters were
		  compared to depositional lithofacies identified by manual
		  core examination. The clusters were ordered by the SOM in a
		  pattern consistent with environmental gradients inferred
		  from core examination: bind/boundstone, grainstone,
		  packstone. and wackestone. This new approach predicted
		  lithofacies identity from well log data with 78.8% accuracy
		  which is more accurate than using a backpropagation neural
		  network (57.3%). The clusters produced by the SOM are
		  ordered with respect to paleoenvironmental energy levels.
		  This energy- related clustering provides geologists and
		  petroleum engineers with valuable geologic information
		  about the logfacies and their interrelationships. This
		  advantage is not obtained in backpropagation neural
		  networks and adaptive resonance theory neural networks.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chang92a,
  author	= {C. -C. Chang and C. -H. Chang and S. -Y. Hwang},
  title		= {A connectionist approach for thresholding},
  booktitle	= {Proc. 11ICPR, International Conference on Pattern
		  Recognition},
  year		= {1992},
  volume	= {III},
  pages		= {522--525},
  organization	= {Int. Assoc Pattern Recognition},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang92b,
  author	= {Chang, W. and Soliman, H. S. and Sung, A. H. },
  title		= {Image data compression using counterpropagation network},
  booktitle	= {1992 IEEE International Conference on Systems, Man and
		  Cybernetics},
  year		= {1992},
  volume	= {1},
  pages		= {405--9},
  organization	= {Dept. of Comput. Sci. , New Mexico Inst. of Min. \&
		  Technol. , Socorro, NM, USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang93a,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Force Directed Self-Organizing Map and its Application to
		  {VLSI} Cell Placement},
  booktitle	= {Proc. ICNN'93, International Conference on Neural
		  Networks},
  year		= {1993},
  volume	= {I},
  pages		= {103--109},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {This paper describes a new self-organizing map called
		  force directed self-organizing map (FDSOM) which can be
		  used in VLSI cell placement with various constraints on
		  their connection and dimension such that the total wire
		  length and area of the resulting placement are minimized.
		  This procedure combines ideas from a force directed
		  relaxation and the self-organization algorithm proposed by
		  Kohonen. It is specially suited for such a
		  self-organization problem that those (input) sample vectors
		  are not easily available. A proof of convergence for this
		  model is given in this paper. With this property, it can
		  therefore be used in CAM or any other computational task.
		  The experimental results obtained in this paper are quite
		  encouraging. All these processes are convergent in a
		  reasonable number of iterations. It was found that the
		  proposed approach is competitive with the state-of-the-art
		  algorithms and uses more fewer nodes and connection
		  weights.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang93b,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Circuit placement in arbitrarily shaped regions using
		  neural network},
  booktitle	= {Proceedings TENCON '93. 1993 IEEE Region 10 Conference on
		  'Computer, Communication, Control and Power Engineering'},
  year		= {1993},
  editor	= {Yuan Baozong},
  volume	= {2},
  pages		= {1150--3},
  organization	= {Dept. of Comput. \& Inf. Sci. , Nat. Chiao Tung Univ. ,
		  Hsinchu, Taiwan},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang93c,
  author	= {Chang, W. and Soliman, H. S. and Sung, A. H. },
  title		= {Preserving visual perception by learning natural
		  clustering},
  booktitle	= {1993 IEEE International Conference on Neural Networks},
  year		= {1993},
  volume	= {2},
  pages		= {661--6},
  organization	= {Dept. of Comput. Sci. , New Mexico Inst. of Min. \&
		  Technol. , Socorro, NM, USA},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang93d,
  author	= {Chang, R. -I. and Hsiao, P. -Y. },
  title		= {Arbitrarily sized cell placement by \mbox{self-organizing}
		  neural networks},
  booktitle	= {Proceedings of the 1993 IEEE International Symposium on
		  Circuits and Systems},
  year		= {1993},
  volume	= {3},
  pages		= {2043--6},
  organization	= {Dept. of Comput. \& Inf. Sci. , Nat. Chiao Tung Univ. ,
		  Hsinchu, Taiwan},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94a,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Force Directed Self-Organizing Maps for {L}-Shaped Cell
		  Placement using Delta Learning Rule},
  pages		= {3381--3386},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, cell placement, modification},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94b,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Artificial Texture Generation using Force Directed
		  Self-Organizing Maps},
  pages		= {4123--4128},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, texture generation},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94c,
  author	= {W. Chang and H. S. Soliman and A. H. Sung},
  title		= {A Vector Quantization Neural Network to Compress Still
		  Monochromatic Images},
  pages		= {4163--4168},
  booktitle	= {Proc. ICNN'94, International Conference on Neural
		  Networks},
  year		= {1994},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  annote	= {application, vector quantization, comparison},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94d,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Fast {S}elf-{O}rganization by Query-Based Algorithm and
		  its Applications},
  booktitle	= {Proc. 1994 Int. Symp. on Speech, Image Processing and
		  Neural Networks},
  year		= {1994},
  volume	= {I},
  pages		= {85--88},
  organization	= {{IEEE} Hong Kong Chapt. of Signal Processing},
  address	= {Hong Kong},
  annote	= {modification},
  dbinsdate	= {oldtimer}
}

@Article{	  chang94e,
  author	= {Chen-Huei Chang and Shu-Yuen Hwang},
  title		= {{2-D} curve partitioning by {K}ohonen feature maps},
  journal	= {Journal of Visual Communication and Image Representation},
  year		= {1994},
  volume	= {5},
  number	= {2},
  pages		= {148--55},
  month		= {June},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94f,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Rectangular {VLSI} cell placement using force directed
		  \mbox{self-organizing} maps and delta learning rules},
  booktitle	= {Proceedings of 1994 IEEE Region 10's Ninth Annual
		  International Conference. Theme: Frontiers of Computer
		  Technology},
  year		= {1994},
  editor	= {Chan, T. K. },
  volume	= {2},
  pages		= {1020--4},
  organization	= {Dept. of Comput. \& Inf. Sci. , Nat. Chiao Tung Univ. ,
		  Hsinchu, Taiwan},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94g,
  author	= {Chang, W. and Soliman, H. S. and Sung, A. H. },
  title		= {Fingerprint image compression by a natural clustering
		  neural network},
  booktitle	= {Proceedings ICIP-94},
  year		= {1994},
  volume	= {2},
  pages		= {341--5},
  organization	= {New Mexico Inst. of Min. \& Technol. , Socorro, NM, USA},
  publisher	= {IEEE Computer Society Press},
  address	= {Los Alamitos, CA, USA},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang94h,
  author	= {Kuo-Chu Chang and Yi-Chuan Lu},
  title		= {Feedback learning: a hybrid {SOFM}/{LVQ} approach for
		  radar target classification},
  booktitle	= {1994 International Symposium on Artificial Neural
		  Networks. ISANN '94. Proceedings},
  year		= {1994},
  pages		= {465--70},
  organization	= {Center of Excellence in Command, Control, Commun. \&
		  Intelligence, George Mason Univ. , Fairfax, VA, USA},
  publisher	= {Nat. Cheng Kung Univ},
  address	= {Tainan, Taiwan},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chang94i,
  author	= {W. Chang and H. S. Soliman and A. H. Sung},
  title		= {Fingerprint image compression by a clustering learning
		  network},
  booktitle	= {Industrial and Engineering Applications of Artificial
		  Intelligence and Expert Systems. Proceedings of the Seventh
		  International Conference},
  publisher	= {Gordon \& Breach},
  year		= {1994},
  editor	= {F. D. Anger and R. V. Rodriguez and M. Ali},
  address	= {Yverdon les Bains, Switzerland},
  pages		= {51--6},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chang95a,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Unsupervised Query-Based Learning Algorithm and It's
		  Application to {K}ohonen's Self-Organizing Maps},
  volume	= {V},
  pages		= {2610--2614},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  abstract	= {Query-based learning (QBL) algorithms have been introduced
		  to supervised multilayer perceptrons to provide good
		  training results for correct classification. Since their
		  queried data are produced by external supervisors,
		  conventional QBL methods can not be directly applied to
		  unsupervised learning models which have no external
		  supervisor exist. In this paper, a novel unsupervised QBL
		  (UQBL) algorithm is proposed. In which, network's training
		  samples are further improved by both the goal-oriented
		  selective attention and the self-regulation property. The
		  proposed UQBL method considers not only the external
		  stimulus but also the internal desire. It is not an
		  anthropomorphic style that overestimates the importance of
		  internal desire. We just try to combine two different
		  system parameters for network training. This method is
		  different from the conventional supervised/unsupervised
		  learning algorithms. Our experiments show that the proposed
		  UQBL method can be successfully applied for Kohonen's self
		  organizing maps (SOM). It can provide faster convergence
		  and is more insensitive to network initialization than the
		  standard {SOM}.},
  dbinsdate	= {oldtimer}
}

@Article{	  chang95b,
  author	= {Chang, Kuo-Chu and Lu, Yi-Chuan},
  title		= {High resolution polarimetric {SAR} target classification
		  with neural network},
  journal	= {IEEE International Conference on Fuzzy Systems},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1995},
  number	= {},
  volume	= {3},
  pages		= {1681--1688},
  abstract	= {An improved version of the SOFM/LVQ classifier currently
		  used in an ATR system for {SAR} imagery is presented. This
		  classifier was originally designed to construct a few
		  number of templates to represent a set of targets with
		  different orientations. The classifier accepts an input
		  target data, computes distances of this data with those
		  representative templates, and then classifies this data to
		  the target class with the shortest distance. with this
		  distance discriminator, a good classification performance
		  was obtained when only target data were tested. However,
		  the simple distance measure produces poor classification
		  results when unknown targets such as natural or manmade
		  clutters are present and when each target is represented by
		  a small number of templates. We correct this deficiency by
		  incorporating an entropy measure into the original
		  classifier. With this entropy discriminator, our system
		  rejects a majority of the false alarms while maintaining a
		  high correct classification rate with a relatively few
		  templates for each target.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chang96a,
  author	= {Chir-Ho Chang and Hsien-Hui Tseng and Bor-Yao Huang},
  title		= {Noise immunization of a neural fuzzy intelligent
		  recognition system by the use of feature and rule
		  extraction technique},
  booktitle	= {Soft Computing in Intelligent Systems and Information
		  Processing. Proceedings of the 1996 Asian Fuzzy Systems
		  Symposium},
  publisher	= {IEEE},
  year		= {1996},
  editor	= {Y-Y Chen and K. Hirota and J-J Yen},
  address	= {New York, NY, USA},
  pages		= {73--8},
  dbinsdate	= {oldtimer}
}

@Article{	  chang97a,
  author	= {W. Chang and H. S. Soliman},
  title		= {Image coding by a neural net classification process},
  journal	= {Applied Artificial Intelligence},
  year		= {1997},
  volume	= {11},
  number	= {1},
  pages		= {33--57},
  abstract	= {A self-organizing neural network performing learning
		  vector quantization (LVQ) to compress image data is
		  proposed. By using unsupervised learning, our LVQ neural
		  model approximates optimal pattern clustering from training
		  images through a memory adaptation process, and builds a
		  compression codebook in the synaptic weight matrix. The
		  neural codebook, trained by example pictures, can be used
		  as a codec to compress and decompress other pictures in a
		  speedy fashion. Special image types, such as fingerprints,
		  verify this property in our experiments. Our approach is
		  compared with other recently developed neural VQ models
		  (neural gas, growing cell structures, and conscious
		  competitive learning) and methodological premises are
		  discussed. The performance of our model is also compared
		  with JPEG and wavelet methods. Other advantages of our
		  neural codec model are its low training time, high
		  utilization of neurons, robust clustering capability, and
		  simple computation. Further, our model has some filtering
		  effects through special training methods and learning
		  enhancement techniques for obtaining a compact neural
		  codebook to yield both high compression and high picture
		  quality.},
  dbinsdate	= {oldtimer}
}

@Article{	  chang97b,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {{VLSI} Circuit Placement with Rectilinear Modules using
		  Three-Layer Force-Directed Self-Organizing Maps},
  journal	= {IEEE Transactions on Neural Networks},
  year		= 1997,
  volume	= 8,
  pages		= {1049--1064},
  dbinsdate	= {oldtimer}
}

@Article{	  chang97c,
  author	= {Ray-I Chang and Pei-Yung Hsiao},
  title		= {Unsupervised Query-Based Learning of Neural Networks Using
		  Selective Attention and Self-Regulation},
  journal	= {IEEE Transactions on Neural Networks},
  year		= 1997,
  volume	= 8,
  pages		= {205--217},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chang98a,
  author	= {Jyh-Shan Chang and J. -H. J. Lin and Tzi-Dar Chiueh},
  title		= {Color image vector quantization using binary tree
		  structured \mbox{self-organizing} feature maps},
  booktitle	= {1998 IEEE International Joint Conference on Neural
		  Networks Proceedings. IEEE World Congress on Computational
		  Intelligence},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {2},
  address	= {New York, NY, USA},
  pages		= {1428--32},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chang98b,
  author	= {Maiga Chang and Horng-Jyh Yu and Jia-Sheng Heh},
  title		= {Evolutionary \mbox{self-organizing} map},
  booktitle	= {1998 IEEE International Joint Conference on Neural
		  Networks Proceedings. IEEE World Congress on Computational
		  Intelligence},
  publisher	= {IEEE},
  year		= {1998},
  volume	= {1},
  address	= {New York, NY, USA},
  pages		= {680--5},
  dbinsdate	= {oldtimer}
}

@Article{	  changchien01a,
  author	= {Changchien, S. W. and Lu, T. -C.},
  title		= {Mining association rules procedure to support on-line
		  recommendation by customers and products fragmentation},
  journal	= {Expert Systems with Applications},
  year		= {2001},
  volume	= {20},
  number	= {4},
  month		= {May 2001},
  pages		= {325--335},
  organization	= {Department of Information Management, Chaoyang University
		  of Technology},
  publisher	= {},
  address	= {},
  abstract	= {Electronic Commerce (EC) has offered a new channel for
		  instant on-line shopping. However, there are too many
		  various products available from a great number of virtual
		  stores on the Internet for Internet shoppers to select.
		  On-line one-to-one marketing therefore becomes a great
		  assistance to Internet shoppers. One of the most important
		  marketing resources is the prior daily transaction records
		  in the database. The great amount of data not only gives
		  the statistics, but also offers the resource of experiences
		  and knowledge. It is quite natural that marketing managers
		  can perform data mining on the daily transactions and treat
		  the shoppers the way they prefer. However, the data mining
		  on a significant amount of transaction records requires
		  efficient tools. Data mining from automatic or
		  semi-automatic exploration and analysis on a large amount
		  of data items set in a database can discover significant
		  patterns and rules underlying the database. The knowledge
		  can be equipped in the on-line marketing system to promote
		  Internet sales. The purpose of this paper is to develop a
		  mining association rules procedure from a database to
		  support on-line recommendation. By customers and products
		  fragmentation, product recommendation based on the hidden
		  habits of customers in the database is therefore very
		  meaningful. The proposed data mining procedure consists of
		  two essential modules. One is a clustering module based on
		  a neural network, Self-Organization Map (SOM), which
		  performs affinity grouping tasks on a large amount of
		  database records. The other rule is extraction module
		  employing rough set theory that can extract association
		  rules for each homogeneous cluster of data records and the
		  relationships between different clusters. The implemented
		  system was applied to a sample of sales records from a
		  database for illustration. },
  dbinsdate	= {2002/1}
}

@InCollection{	  chantelou96a,
  author	= {D. Chantelou and G. Hebrail and C. Muller},
  title		= {Visualizing 2665 electric power load curves an a single
		  {A4} sheet of paper},
  booktitle	= {ISAP `96. International Conference on Intelligent Systems
		  Applications to Power Systems Proceedings},
  publisher	= {IEEE},
  year		= {1996},
  editor	= {O. A. Mohammed and K. Tomsovic},
  address	= {New York, NY, USA},
  pages		= {126--32},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chao92a,
  author	= {Chao, J. and Minowa, K. and Tsujii, S. },
  title		= {Unsupervised learning of {3D} objects conserving global
		  topological order},
  booktitle	= {IEEE International Conference on Systems Engineering},
  year		= {1992},
  pages		= {24--7},
  organization	= {Dept. of Electr. \& Electron. Eng. , Chuo Univ. , Toyko,
		  Japan},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@Article{	  chao93a,
  author	= {Chao, J. and Minowa, K. and Tsujii, S. },
  title		= {Unsupervised learning of {3D} objects conserving global
		  topological order},
  journal	= {IEICE Transactions on Fundamentals of Electronics,
		  Communications and Computer Sciences},
  year		= {1993},
  volume	= {E76-A},
  number	= {5},
  pages		= {749--53},
  month		= {May},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chao94a,
  author	= {Jinhui Chao and Kenji Minowa and Shigeo Tsujii},
  title		= {Acquistion of Global Topology for {3D} Objects with Local
		  Competition},
  booktitle	= {Proc. ICANN'94, International Conference on Artificial
		  Neural Networks},
  year		= {1994},
  editor	= {Maria Marinaro and Pietro G. Morasso},
  volume	= {II},
  pages		= {1460--1463},
  publisher	= {Springer},
  address	= {London, UK},
  annote	= {application, extension, internal representations},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chao94b,
  author	= {Chao, J. and Nakayama, J. and Tsujii, S. },
  title		= {Acquisition of global topology for {3D} objects with local
		  competition},
  booktitle	= {APCCAS `94. 1994 IEEE Asia-Pacific Conference on Circuits
		  and Systems},
  year		= {1994},
  pages		= {673--7},
  organization	= {Dept. of Electr. \& Electron. Eng. , Chuo Univ. , Tokyo,
		  Japan},
  publisher	= {IEEE},
  address	= {New York, NY, USA},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chao96a,
  author	= {Jinhui Chao and J. Nakayama},
  title		= {Cubical singular simplex model for {3D} objects and fast
		  computation of homology groups},
  booktitle	= {Proceedings of the 13th International Conference on
		  Pattern Recognition},
  publisher	= {IEEE Computer Society Press},
  year		= {1996},
  volume	= {4},
  address	= {Los Alamitos, CA, USA},
  pages		= {190--4},
  dbinsdate	= {oldtimer}
}

@Article{	  chapline97a,
  author	= {George Chapline},
  title		= {Spontaneous origin of topological complexity in
		  \mbox{self-organizing} neural networks},
  journal	= {Network: Computation in Neural Systems},
  year		= 1997,
  volume	= 8,
  pages		= {185--194},
  dbinsdate	= {oldtimer}
}

@Article{	  chappelier01a,
  author	= {Chappelier, J. C and Gori, M. and Grumbach, A.},
  title		= {Time in connectionist models},
  journal	= {Sequence learning. Paradigms, algorithms, and applications
		  (Lecture Notes in Artificial Intelligence Vol.1828).
		  Springer-Verlag, Berlin, Germany; 2001; xiii+387
		  pp.p.105--34},
  year		= {2001},
  volume	= {},
  pages		= {105--34},
  abstract	= {The prototypical use of "classical" connectionist models
		  (including the multilayer perceptron (MLP), the Hopfield
		  network and the Kohonen self-organizing map) concerns
		  static data processing. These classical models are not well
		  suited to working with data varying over time. In response
		  to this, temporal connectionist models have appeared and
		  constitute a continuously growing research field. The
		  purpose of this chapter is to present the main aspects of
		  this research area and to review the key connectionist
		  architectures that have been designed for solving temporal
		  problems. We present the fundamentals of temporal
		  processing with neural networks. Several temporal
		  connectionist models are then detailed. As a matter of
		  illustration, important applications are reviewed. We
		  conclude with the presentation of a promising future issue:
		  the extension of temporal processing to even more complex
		  structured data.},
  dbinsdate	= {2002/1}
}

@InCollection{	  chappelier96a,
  author	= {J. -C. Chappelier and A. Grumbach},
  title		= {A {K}ohonen map for temporal sequences},
  booktitle	= {Neural Networks and Their Applications. Conference
		  Proceedings},
  publisher	= {Domaine Univ. Saint-Jerome},
  year		= {1996},
  address	= {Marseille, France},
  pages		= {104--10},
  dbinsdate	= {oldtimer}
}

@Article{	  chappell93a,
  author	= {Geoffrey J. Chappell and John G. Taylor},
  title		= {The Temporal {K}ohonen Map},
  journal	= {Neural Networks},
  year		= {1993},
  volume	= {6},
  pages		= {441--445},
  dbinsdate	= {oldtimer}
}

@Article{	  charalambous00a,
  author	= {Charalambous, C. and Charitou, A. and Kaourou, F.},
  title		= {Comparative analysis of artificial neural network models:
		  Application in bankruptcy prediction},
  journal	= {ANNALS OF OPERATIONS RESEARCH},
  year		= {2000},
  volume	= {99},
  pages		= {403--425},
  abstract	= {This study compares the predictive performance of three
		  neural network methods, namely the learning vector
		  quantization, the radial basis function, and the
		  feedforward network that uses the conjugate gradient
		  optimization algorithm, with the performance of the
		  logistic regression and the backpropagation algorithm. All
		  these methods are applied to a dataset of 139 matched-pairs
		  of bankrupt and non-bankrupt US firms for the period
		  1983--1994. The results of this study indicate that the
		  contemporary neural network methods applied in the present
		  study provide superior results to those obtained from the
		  logistic regression method and the backpropagation
		  algorithm.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  charalambous01a,
  author	= {Charalambous, C. and Hadjinicola, G. C. and Muller, E.},
  title		= {Product positioning using principles from the
		  self-organizing map},
  booktitle	= {ARTIFICAL NEURAL NETWORKS-ICANN 2001, PROCEEDINGS},
  year		= {2001},
  pages		= {457--463},
  abstract	= {This paper presents a methodology that identifies the
		  position of a new product in the attribute space. The
		  methodology uses principles from Kohonen's self-organizing
		  feature map. The algorithm presented is robust and can be
		  used for a number of objective functions commonly used in
		  the product positioning problem. The method can also be
		  used in competitive environments where other competing
		  products are already present in the market. Furthermore,
		  the algorithm can accommodate single-choice models (the
		  consumer purchases the product "closest" to his/her
		  preferences) and probabilistic-choice models (the consumer
		  assigns to each product a probability for purchasing it).},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chaudhary92a,
  author	= {Chaudhary, S. D. and Kalra, P. K. and Srivastava, S. C. },
  title		= {Short term electric load forecasting using artificial
		  neural network},
  booktitle	= {Expert System Application to Power Systems IV
		  Proceedings},
  year		= {1992},
  editor	= {Dillon, T. S. },
  pages		= {159--63},
  organization	= {Dept. of Electr. Eng. , Indian Inst. of Technol. , India},
  publisher	= {CRL Publishing},
  address	= {Aldershot, UK},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chaudhuri95a,
  author	= {Chaudhuri, T. R. and Yeh, J. C. -H. and Hamey, L. G. C.
		  and Westcott, C. T. },
  title		= {Baked product classification with the use of a
		  \mbox{self-organising} map},
  booktitle	= {Proceedings of the Sixth Australian Conference on Neural
		  Networks (ACNN`95)},
  year		= {1995},
  editor	= {Charles, M. and Latimer, C. },
  pages		= {152--5},
  organization	= {Dept. of Comput. , Macquarie Univ. , North Ryde, NSW,
		  Australia},
  publisher	= {Univ. Sydney},
  address	= {Sydney, NSW, Australia},
  dbinsdate	= {oldtimer}
}

@Article{	  chauhan97a,
  author	= {S. Chauhan and M. P. Dave},
  title		= {{K}ohonen neural network classifier for voltage collapse
		  margin estimation},
  journal	= {Electric Machines and Power Systems},
  year		= {1997},
  volume	= {25},
  number	= {6},
  pages		= {607--19},
  dbinsdate	= {oldtimer}
}

@Article{	  chebira97a,
  author	= {A. Chebira and K. Madani and G. Mercier},
  title		= {Various ways for building a multi-neural network system:
		  application to a control process},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1997},
  volume	= {3077},
  pages		= {148--59},
  note		= {(Applications and Science of Artificial Neural Networks
		  III Conf. Date: 21--24 April 1997 Conf. Loc: Orlando, FL,
		  USA Conf. Sponsor: SPIE)},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chedid94a,
  author	= {R. Chedid and N. Najjar and F. Chedid},
  title		= {A neural network approach for finite element software},
  booktitle	= {Proceedings of the IASTED International Conference:
		  Modelling, Simulation and Identification},
  publisher	= {IASTED},
  year		= {1994},
  editor	= {Y. Kagawa},
  address	= {Calgary, Alta. , Canada},
  pages		= {232--6},
  dbinsdate	= {oldtimer}
}

@Article{	  chedid96a,
  author	= {R. Chedid and N. Najjar},
  title		= {Automatic finite-element mesh generation using artificial
		  neural networks---Part {I}: Prediction of mesh density},
  journal	= {IEEE Transactions on Magnetics},
  year		= {1996},
  volume	= {32},
  number	= {5, pt. 3},
  pages		= {5173--8},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chella01a,
  author	= {Chella, A. and Guarino, M. D. and Pirrone, R.},
  title		= {A {SOM}/{ARSOM} hierarchy for the description of dynamic
		  scenes},
  booktitle	= {AI*IA 2001: Advances in Artificial Intelligence. 7th
		  Congress of the Italian Association for Artificial
		  Intelligence. Proceedings (Lecture Notes in Artificial
		  Intelligence Vol.2175). Springer-Verlag, Berlin, Germany},
  year		= {2001},
  volume	= {},
  pages		= {362--8},
  abstract	= {A neural architecture is presented, aimed at describing
		  the dynamic evolution of complex structures inside a video
		  sequence. The proposed system is arranged as a tree of
		  self-organizing maps. Leaf nodes are implemented by ARSOM
		  networks as a way to code dynamic inputs, while classical
		  SOMs are used to implement the upper levels of the
		  hierarchy. Depending on the application domain, inputs are
		  made by suitable low level features extracted frame by
		  frame of the sequence. Theoretical foundations of the
		  architecture are reported along with a detailed outline of
		  its structure, and encouraging experimental results.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chella01b,
  author	= {Chella, A. and Guarino, M. D. and Pirrone, R.},
  title		= {Description of dynamic structured scenes by a
		  {SOM}/{ARSOM} hierarchy},
  booktitle	= {Artificial Neural Networks---ICANN 2001. International
		  Conference. Proceedings (Lecture Notes in Computer Science
		  Vol.2130). Springer-Verlag, Berlin, Germany},
  year		= {2001},
  volume	= {},
  pages		= {1034--41},
  abstract	= {A neural architecture is presented whose intention is to
		  describe the dynamic evolution of complex structures inside
		  a video sequence. The proposed system is arranged as a tree
		  of self-organizing maps (SOMs). Leaf nodes are implemented
		  by ARSOM (autoregressive SOM) networks as a way to code
		  dynamic inputs, while classical SOMs are used to implement
		  the upper levels of the hierarchy. Depending on the
		  application domain, inputs are made by suitable low-level
		  features extracted frame-by-frame from the sequence.
		  Theoretical foundations of the architecture are reported,
		  along with a detailed outline of its structure, and
		  encouraging experimental results are obtained.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chella91a,
  author	= {Chella, A. and Gioiello, M. and Sorbello, F. },
  title		= {A new digital architecture implementing the {K}ohonen
		  maps},
  booktitle	= {Digital Signal Processing---91. Proceedings of the
		  International Conference},
  year		= {1991},
  editor	= {Cappellini, V. and Constantinides, A. G. },
  pages		= {514--19},
  organization	= {Dipartimento di Ingegneria Elettrica, Palermo Univ. ,
		  Italy},
  publisher	= {Elsevier},
  address	= {Amsterdam, Netherlands},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chella97a,
  author	= {A. Chella and S. Gaglio and V. Mulia and G. Sajeva},
  title		= {An {ASSOM} neural network to represent actions performed
		  by an autonomous agent},
  booktitle	= {Artificial Neural Networks---ICANN '97. 7th International
		  Conference Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1997},
  editor	= {W. Gerstner and A. Germond and M. Hasler and J. -D.
		  Nicoud},
  address	= {Berlin, Germany},
  pages		= {799--804},
  dbinsdate	= {oldtimer}
}

@Article{	  chen00a,
  author	= {Dar Ren Chen and Ruey Feng Chang and Yu Len Huang},
  title		= {Breast cancer diagnosis using \mbox{self-organizing} map
		  for sonography},
  journal	= {Ultrasound in Medicine and Biology},
  year		= {2000},
  volume	= {26},
  pages		= {405--11},
  abstract	= {The purpose of this study was to evaluate the performance
		  of neural network model self-organizing maps (SOM) in the
		  classification of benign and malignant sonographic breast
		  lesions. A total of 243 breast tumors (82 malignant and 161
		  benign) were retrospectively evaluated. When a sonogram was
		  performed, the analog video signal was captured to obtain a
		  digitized sonographic image. The physician selected the
		  region of interest in the sonography. An {SOM} model using
		  24 autocorrelation texture features classified the tumor as
		  benign or malignant. In the experiment, cases were sampled
		  with k-fold cross-validation (k=10) to evaluate the
		  performance using receiver operating characteristic (ROC)
		  curves. The ROC area index for the proposed {SOM} system is
		  0.9357+or-0.0152, the accuracy is 85.6%, the sensitivity is
		  97.6%, the specificity is 79.5%, the positive predictive
		  value is 70.8%, and the negative predictive value is 98.5%.
		  This computer-aided diagnosis system can provide a useful
		  tool and its high negative predictive value could
		  potentially help avert benign biopsies.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen00b,
  author	= {Chen, C. H. and Shrestha, Binesh},
  title		= {Classification of multi-sensor remote sensing images using
		  Self-Organizing Feature Maps and radial basis function
		  networks},
  booktitle	= {International Geoscience and Remote Sensing Symposium
		  (IGARSS)},
  year		= {2000},
  editor	= {},
  volume	= {2},
  pages		= {711--713},
  organization	= {Univ of Massachusetts Dartmouth},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {Neural networks have been known for its classification
		  capability and have been used quite extensively in pixel
		  classification for remote sensing in recent years. In this
		  paper we propose the use of Self-Organizing Feature Map
		  (SOFM) neural networks to evaluate the centers for the
		  hidden neurons in RBF neural networks for pixel
		  classification. The combined use of both neural networks
		  leads to a much better classification than the use of only
		  one of the two networks. For the experimental study the
		  image data considered is from both SAR and ATM sensors and
		  for each pixel the combined data form a 15 dimensional
		  feature vector. Five pattern classes are defined for 5
		  crops. The use of SOFM alone provides only 62.7% correct at
		  best. If only RBF network is used the best reported
		  performance is 89.5% correct. However by combining SOFM and
		  RBF, the best average performance is 95.15% correct, a
		  dramatic improvement over the use of either network.
		  Performance degradation is about 5% if we use only the SAR
		  data. This points to the needs for the data level fusion
		  for most effective pixel classification. Performance
		  comparison is also made with the traditional k-nearest
		  neighbor decision rule which provides 86.5 correct. The
		  results clearly demonstrate the advantage of multi-sensor
		  remote sensing using neural networks.},
  dbinsdate	= {2002/1}
}

@Article{	  chen00c,
  author	= {Chen, Fei-Long and Liu, Shu-Fan},
  title		= {Neural-network approach to recognize defect spatial
		  pattern in semiconductor fabrication},
  journal	= {IEEE Transactions on Semiconductor Manufacturing},
  year		= {2000},
  volume	= {13},
  number	= {3},
  month		= {Aug},
  pages		= {366--373},
  organization	= {Natl Tsing Hua Univ},
  publisher	= {IEEE},
  address	= {Piscataway, NJ},
  abstract	= {Yield enhancement in semiconductor fabrication is
		  important. Even though IC yield loss may be attributed to
		  many problems, the existence of defects on the wafer is one
		  of the main causes. When the defects on the wafer form
		  spatial patterns, it is usually a clue for the
		  identification of equipment problems or process variations.
		  This research intends to develop an intelligent system,
		  which will recognize defect spatial patterns to aid in the
		  diagnosis of failure causes. The neural-network
		  architecture named adaptive resonance theory network 1
		  (ART1) was adopted for this purpose. Actual data obtained
		  from a semiconductor manufacturing company in Taiwan were
		  used in experiments with the proposed system. Comparison
		  between ART1 and another unsupervised neural network,
		  self-organizing map (SOM), was also conducted. The results
		  show that ART1 architecture can recognize the similar
		  defect spatial patterns more easily and correctly.},
  dbinsdate	= {2002/1}
}

@Article{	  chen00d,
  author	= {Yinyong Chen and Reggia, J. A.},
  title		= {The temporal correlation hypothesis for self-organizing
		  feature maps},
  journal	= {International-Journal-of-Systems-Science},
  year		= {2000},
  volume	= {31},
  pages		= {911--21},
  abstract	= {Feature maps, in which one or more aspects of the
		  environment are systematically represented over the surface
		  of the cerebral cortex, are often found in primary sensory
		  and motor cortical regions of the vertebrate brain. They
		  have inspired a great deal of computational modelling, and
		  this has provided evidence that such maps are emergent
		  properties of the interactions of numerous cortical neurons
		  and their adaptive, nonlinear connections. We address the
		  issue of how multiple feature maps that coexist in the same
		  region of the cerebral cortex align with each other. We
		  hypothesize that such alignment is governed by temporal
		  correlations: features in one map that are temporally
		  correlated with those in another come to occupy the same
		  spatial locations over time. To examine the feasibility of
		  this hypothesis and to establish some of its detailed
		  implications, we initially studied a computational model of
		  the primary sensori-motor cortex. Coexisting sensory and
		  motor maps formed and generally aligned in a fashion
		  consistent with the temporal correlation hypothesis. We
		  summarize these results, and then mathematically analyse a
		  simplified model of self-organization during unsupervised
		  learning. We show that the properties observed
		  computationally are quite general: that temporally
		  correlated inputs become spatially correlated (i.e.
		  aligned), while input patterns that are temporally
		  anti-correlated tend to result in mutually exclusive (i.e.
		  unaligned) spatial distributions. This work provides a
		  framework in which to interpret and understand future
		  experimental studies of map relationships.},
  dbinsdate	= {2002/1},
  merjanote     = {Last name checked from internet}
}

@Article{	  chen01a,
  author	= {Chen, H. B. and Grant-Muller, S. and Mussone, L. and
		  Montgomery, F.},
  title		= {A study of hybrid neural network approaches and the
		  effects of missing data on traffic forecasting},
  journal	= {NEURAL COMPUTING \& APPLICATIONS},
  year		= {2001},
  volume	= {10},
  number	= {3},
  pages		= {277--286},
  abstract	= {In this pal)er we present an application of hybrid neural
		  network approaches and an assessment of the effects of
		  missing data on motorway traffic flow forecasting. Two 1
		  ybrid aj)13roaches are developed using a Self-Organising
		  Map (SOM) to initially, classify traffic into different
		  states. The first hybrid approach includes four
		  Auto-Regressive Integrated Moving Average (ARIMA) models,
		  whilst the second uses two Multi-Layer Perception (MLP)
		  models. It was found that the SOM/ARIMA hybrid approach
		  out-performs all individual ARIMA models, whilst the
		  SOM/MLP hybrid approach achieves superior forecasting
		  performance to all models used in this study, including
		  three naive models. The effects of different proportions of
		  missing data on Neural Network (NN) performance when
		  forecasting traffic flow are assessed and several initial
		  substitution options to replace missing data are discussed.
		  Overall, it is shown that ARIMA models are more sensitive
		  to the percentage of missing data than neural networks in
		  this context.},
  dbinsdate	= {2002/1}
}

@Article{	  chen01b,
  author	= {Chen, J. J. W. and Peck, K. and Hong, T. M. and Yang, S.
		  C. and Sher, YP and Shih, JY and Wu, R and Cheng, JL and
		  Roffler, SR and Wu, CW and Yang, PC},
  title		= {Global analysis of gene expression in invasion by a lung
		  cancer model},
  journal	= {CANCER RESEARCH},
  year		= {2001},
  volume	= {61},
  number	= {13},
  month		= {JUL 1},
  pages		= {5223--5230},
  abstract	= {Metastasis is a complicated multistep process that
		  involves interactions between cancer cells and their
		  surrounding microenvironments. Previously, we have
		  established a series of lung adenocarcinoma cell lines with
		  varying degrees of invasiveness. Tracheal graft assay
		  confirmed that cell lines with higher in vitro invasiveness
		  had greater in vivo invasive potential. In this study, me
		  used these model cell lines to identify invasion associated
		  genes using cDNA microarray with colorimetric detection. A
		  more invasive subline, CL 1--5-F 4, derived from metastatic
		  lung tumor of severe combined immunodeficient mice
		  inoculated with CL 1--5 cells, was combined with CL 1--0,
		  CL 1--1, and CL 1--5 in cDNA microarray screening. cDNA
		  microarray membranes, each containing 9600 nonredundant
		  expressed sequence tag clones, mere used to identify
		  differentially expressed genes in these cell lines. For
		  statistical analysis, self-organizing map algorithm was
		  performed to identify the expression patterns. Positive
		  correlation between gene expression levels and cell line
		  invasiveness was found in 2.9% of the 9600 putative genes,
		  On the other hand, negative correlation was found in 3.3%
		  of the genes. The trends of expression of some of the genes
		  were also confirmed by Northern hybridization and flow
		  cytometry, Our data demonstrated that genes related to cell
		  adhesion, motility, angiogenesis, signal transduction, and
		  some other expressed sequence tag genes may play
		  significant roles in the metastasis process. These results
		  substantiate the model system with which one can identify
		  invasion-associated genes by using cDNA microarray and
		  cancer cell lines of different invasiveness. This technique
		  may allow us to explore complex interactions between
		  multiple genes that orchestrate the process of cancer
		  metastasis.},
  dbinsdate	= {2002/1}
}

@Article{	  chen02a,
  author	= {Chen, J. and Hagiwara, I. and Su, X. and Shi, Q. Z.},
  title		= {A bispectrum feature extraction enhanced structure damage
		  detection approach},
  journal	= {JSME INTERNATIONAL JOURNAL SERIES C-MECHANICAL SYSTEMS
		  MACHINE ELEMENTS AND MANUFACTURING},
  year		= {2002},
  volume	= {45},
  number	= {1},
  month		= {MAR},
  pages		= {121--126},
  abstract	= {The subject of structure defect diagnosis has been
		  extensively investigated in the field of nondestructive
		  testing (NDT). In this paper, a new approach for detecting
		  structure damage is proposed, which is based on the
		  combination of the bispectrum feature extraction method and
		  the learning vector quantization (LVQ) identification
		  method. Because bispectrum analysis possesses the
		  capability of restraining Gaussian noise, it may be
		  employed to enhance the performance of the feature
		  extraction method. In simulation, by using the proposed
		  method, it has been shown that very high accuracy of
		  structure damage identification can be obtained compared
		  with the modal assurance criterion (MAC) formed from the
		  modal parameters method, especially in the case of low
		  signal-to-noise ratio environment.},
  dbinsdate	= {2002/1}
}

@InProceedings{	  chen90a,
  author	= {Daowen Chen and Yuqing Gao},
  title		= {Classification and trajectory for {C}hinese speech by
		  self-organization feature maps},
  booktitle	= {Proc. INNC'90, Int. Neural Network Conference},
  year		= {1990},
  volume	= {I},
  pages		= {195},
  organization	= {Thomsom; SUN; British Computer Society ; et al},
  publisher	= {Kluwer},
  address	= {Dordrecht, Netherlands},
  dbinsdate	= {oldtimer}
}

@Article{	  chen92a,
  author	= {M. -S. Chen and Hsiao-Chuan Wang},
  title		= {A decision enhanced pattern classifier based on neural
		  network approach},
  journal	= {Pattern Recognition Letters},
  year		= {1992},
  volume	= {13},
  number	= {5},
  pages		= {315--323},
  month		= {May},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen92b,
  author	= {Oscal T. -C. Chen and Bing J. Sheu and Wai-Chi Fang},
  title		= {Adaptive Vector Quantization for Image Compression Using
		  Self-Organization Approach},
  booktitle	= {Proc. ICASSP-92, International Conference on Acoustics,
		  Speech and Signal processing},
  year		= {1992},
  volume	= {II},
  pages		= {385--388},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@Article{	  chen93a,
  title		= {Artificial neural networks and their applications in
		  control and system engineering: an introduction of neural
		  networks},
  author	= {Yunping Chen},
  journal	= {Power System Technology},
  year		= {1993},
  colume	= {1},
  pages		= {56--58},
  month		= {January},
  note		= {(in Chinese)},
  annote	= {General review},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen93b,
  author	= {X. Chen and R. Kothari and P. Klinkhachorn},
  title		= {Reduced Color Image Based on Adaptive Palette Color
		  Selection Using Neural Networks},
  booktitle	= {Proc. WCNN'93, World Congress on Neural Networks},
  year		= {1993},
  volume	= {I},
  pages		= {555--558},
  organization	= {{INNS}},
  publisher	= {Lawrence Erlbaum},
  address	= {Hillsdale},
  dbinsdate	= {oldtimer}
}

@Article{	  chen94a,
  author	= {Yung-Sheng Chen and Yu-Chang Hsu},
  title		= {Image segmentation of a color-blindness plate},
  journal	= {Applied Optics},
  year		= {1994},
  volume	= {33},
  number	= {29},
  pages		= {6818--22},
  month		= {Oct},
  abstract	= {Segmentation of a color-blindness plate (CBP) image is
		  discussed and its method is developed. The method uses
		  self-organizing feature map algorithm and labeling process
		  as well as computation of spatial distance. Experimental
		  results show that the method can classify the CBP image and
		  background successfully.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen95a,
  author	= {Yifeng Chen and Yuanda Cao},
  title		= {A Hybrid Neural Network for Spatio-Temporal Pattern
		  Recognition},
  volume	= {III},
  pages		= {1414--1417},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen95b,
  author	= {Yifeng Chen and Zhuoqun Xu},
  title		= {A High-Dimensional {SOFM} Vector Quantizer With Weightless
		  Neural Prediction},
  volume	= {III},
  pages		= {1418--1421},
  booktitle	= {Proc. ICNN'95, IEEE International Conference on Neural
		  Networks},
  year		= {1995},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen95c,
  author	= {Yifeng Chen},
  title		= {A high-dimensional {SOFM} neural vector quantizer for
		  image compression},
  booktitle	= {Proceedings of International Conference on Neural
		  Information Processing (ICONIP `95)},
  year		= {1995},
  editor	= {Zhong, Y. and Yang, Y. and Wang, M. },
  volume	= {2},
  pages		= {698--702},
  organization	= {Comput. Sci. Dept. , Peking Univ. , Beijing, China},
  publisher	= {Publishing House of Electron. Ind},
  address	= {Beijing, China},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chen95d,
  author	= {Ting-Yu Chen and Jih-Chang Wang and Hsin-Li Chang},
  title		= {Applying habitual domains to modify the
		  \mbox{self-organizing} map},
  booktitle	= {Fuzzy Logic for the Applications to Complex Systems.
		  Proceedings of the International Joint Conference of
		  CFSA/IFIS/SOFT '95 on Fuzzy Theory and Applications},
  publisher	= {World Scientific},
  year		= {1995},
  editor	= {W. L. Chiang and J. Lee},
  address	= {Singapore},
  pages		= {302--7},
  dbinsdate	= {oldtimer}
}

@Article{	  chen96a,
  author	= {Hsinchun Chen and C. Schuffels and R. Orwig},
  title		= {Internet categorization and search: a
		  \mbox{self-organizing} approach},
  journal	= {Journal of Visual Communication and Image Representation},
  year		= {1996},
  volume	= {7},
  number	= {1},
  pages		= {88--102},
  abstract	= {The problems of information overload and vocabulary
		  differences have become more pressing with the emergence of
		  increasingly popular Internet services. The main
		  information retrieval mechanisms provided by the prevailing
		  Internet WWW software are based on either keyword search
		  (e.g., the Lycos server at CMU, the Yahoo server at
		  Stanford) or hypertext browsing (e.g., Mosaic and
		  Netscape). This research aims to provide an alternative
		  concept-based categorization and search capability for WWW
		  servers based on selected machine learning algorithms. Our
		  proposed approach, which is grounded on automatic textual
		  analysis of Internet documents (homepages), attempts to
		  address the Internet search problem by first categorizing
		  the content of Internet documents. We report results of our
		  recent testing of a multilayered neural network clustering
		  algorithm employing the Kohonen self-organizing feature map
		  to categorize (classify) Internet homepages according to
		  their content. The category hierarchies created could serve
		  to partition the vast Internet services into
		  subject-specific categories and databases and improve
		  Internet keyword searching and/or browsing.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  chen96b,
  author	= {O. T. -C. Chen and Chih-Yung Chen and Hwai-Tsu Cheng and
		  Fang-Ru Hsu and Huang-Lin Yang and Youn-Gwo Lee},
  title		= {A multi-lingual speech recognition system using a neural
		  network approach},
  booktitle	= {ICNN 96. The 1996 IEEE International Conference on Neural
		  Networks},
  publisher	= {IEEE},
  year		= {1996},
  volume	= {3},
  address	= {New York, NY, USA},
  pages		= {1576--81},
  dbinsdate	= {oldtimer}
}

@Article{	  chen97a,
  author	= {Yuhai Chen and A. T. Chwang},
  title		= {Self-organized feature map of particle image for flow
		  measurement},
  journal	= {Proceedings of the SPIE---The International Society for
		  Optical Engineering},
  year		= {1997},
  volume	= {3172},
  pages		= {142--52},
  dbinsdate	= {oldtimer}
}

@Article{	  chen98a,
  author	= {Hsinchun Chen and Andrea L. Houston and Robin R. Sewell
		  and Bruce R. Schatz},
  title		= {Internet Browsing and Searching: User Evaluations of
		  Category Map and Concept Space Techniques},
  journal	= {Journal of the American Society for Information Science},
  year		= 1998,
  volume	= {49},
  pages		= {582--603},
  abstract	= {The Internet provides an exceptional testbed for
		  developing algorithms that can improve browsing and
		  searching large information spaces. Browsing and searching
		  tasks are susceptible to problems of information overload
		  and vocabulary differences. Two of these algorithms, a
		  Kohonen algorithm category map for browsing and an
		  automatically generated concept space algorithm for
		  searching, are investigated if they can help improve
		  browsing and/or searching the Internet.},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  chen99a,
  author	= {Chen, Z. and Feng, T. J. and Houkes, Z.},
  title		= {Texture segmentation based on wavelet and {K}ohonen
		  network for remotely sensed images},
  booktitle	= {IEEE SMC'99 Conference Proceedings. 1999 IEEE
		  International Conference on Systems, Man, and Cybernetics.
		  },
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  year		= {1999},
  volume	= {6},
  pages		= {816--21},
  abstract	= {In this paper, an approach based on wavelet decomposition
		  and Kohonen's self-organizing map is developed for image
		  segmentation. After performing the 2D wavelet transform of
		  image, some features are extracted for texture
		  segmentation, and the Kohonen neural network is used to
		  accomplish feature clustering. The experimental results
		  demonstrated the satisfactory effect of the proposed
		  approach both for simulated textured image and
		  multi-spectral remotely sensed images.},
  dbinsdate	= {oldtimer}
}

@Article{	  chen99b,
  author	= {Chen, Zengqiang and He, Jiangfeng and Yuan, Zhuzhi},
  title		= {An adaptive identification and control scheme using radial
		  basis function networks},
  journal	= {Journal of Systems Engineering and Electronics},
  year		= {1999},
  volume	= {10},
  number	= {1},
  pages		= {54--61},
  abstract	= {Adaptive identification and control of nonlinear dynamical
		  systems are investigated using radial basis function
		  networks (RBF). Firstly, a novel approach to train the RBF
		  is introduced, which employs an adaptive fuzzy generalized
		  learning vector quantization (AFGLVQ) technique and
		  recursive least squares algorithm with variable forgetting
		  factor (VRLS). The AFGLVQ adjusts the centers of the RBF
		  while the VRLS updates the connection weights of the
		  network. The identification algorithm has the properties of
		  rapid convergence and persistent adaptability that make it
		  suitable for real-time control. Secondly, on the basis of
		  the one-step ahead RBF predictor, the control law is
		  optimized iteratively through a numerical stable Davidon's
		  least squares-based (SDLS) minimization approach. Four
		  nonlinear examples are simulated to demonstrate the
		  effectiveness of the identification and control
		  algorithms.},
  dbinsdate	= {oldtimer}
}

@Article{	  chen99c,
  author	= {Chen, T. and Chen, L. H. and Ma, K. K.},
  title		= {Colour image indexing using {SOM} for region-of-interest
		  retrieval},
  journal	= {Pattern Analysis and Applications},
  year		= {1999},
  volume	= {2},
  pages		= {164--71},
  abstract	= {We present a novel approach to image indexing by
		  incorporating the Kohonen's self-organising map (SOM) for
		  content-based image retrieval. An important and unique
		  aspect of our interactive scheme is to allow the user to
		  select a region-of-interest from the sample image, and
		  subsequent query concentrates on matching the regional
		  colour features to find images containing similar regions
		  as indicated by the user. The {SOM} algorithm is capable of
		  adaptively partitioning each image into several homogeneous
		  regions for representing and indexing the image. This is
		  achieved by unsupervised clustering and classification of
		  pixel-level features, called local neighbourhood
		  histograms, without a priori knowledge about the data
		  distribution in the feature space. The indexes generated
		  from the resultant prototypes of {SOM} learning demonstrate
		  fairly good performance over an experimental image
		  database, and therefore suggest the effectiveness and
		  significant potential of our proposed indexing and
		  retrieval strategy for application to content-based image
		  retrieval.},
  dbinsdate	= {oldtimer}
}

@InCollection{	  cheneval95a,
  author	= {Y. Cheneval},
  title		= {Packlib, an interactive environment to develop modular
		  software for data processing},
  booktitle	= {From Natural to Artificial Neural Computation.
		  International Workshop on Artificial Neural Networks.
		  Proceedings},
  publisher	= {Springer-Verlag},
  year		= {1995},
  editor	= {J. Mira and F. Sandoval},
  address	= {Berlin, Germany},
  pages		= {673--82},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cheng00a,
  author	= {Guojian Cheng and Andreas Zell},
  title		= {Externally Growing Cell Structures for Pattern
		  Classification},
  booktitle	= {Proceeding of the ICSC Symposia on Neural Computation
		  (NC'2000) May 23-26, 2000 in Berlin, Germany},
  editor	= {H. Bothe and R. Rojas},
  year		= {2000},
  organization	= {University Tubingen; Wilhelm-Schickard-Institut for
		  Computer Science, Dept. Computer Architecture},
  publisher	= {ICSC Academic Press},
  dbinsdate	= {oldtimer}
}



@InProceedings{	  cheng92a,
  author	= {Yizong Cheng},
  title		= {Clustering with Competing Self-Organizing Maps},
  booktitle	= {Proc. IJCNN'92, International Joint Conference on Neural
		  Networks},
  year		= {1992},
  publisher	= {IEEE Service Center},
  address	= {Piscataway, NJ},
  volume	= {IV},
  pages		= {785--790},
  dbinsdate	= {oldtimer}
}

@InProceedings{	  cheng92b,
  author	= {Y. M. Cheng and others},
  title		= {Hybrid Segmental-{LVQ} for Large Vocabulary Speech
		  Recognition},
  booktitle	= {Proc. ICASSP-92, International Conference on Acoustics,
		  Speech and Sig