1. Computationally intelligent hybrid-paradigm environments (HYPE)

1.1 Application of fuzzy adaptive arithmetic in design

Abstract

Results

A generic C++ library based on evolutionary strategies and domain knowledge was developed to solve the layout of a restricted graph (a network of nodes and arcs connecting the nodes) as a planar orthogonal diagram (see Figure 1). Tekla is currently negotiating with us concerning the integration of this tool into their products.

A generic C++ library for the Stratified Quasi Monte Carlo method was developed and applied for the VAR worst case analysis.

Two MSc thesis were written by the project members.

Project information

Participants

• VTT Information Technology
• Tekla Oy
• Trema Oy
• Delisoft Oy

Project dates

Project volume

Project manager

Figure 1. An orthogonal layout for a restricted graph generated by our evolution based system

Publications

Articles in scientific text books

Articles in proceedings

E. Hyvönen, S. De Pascale, Interval Constraint Generalization of MS Excel. Proceedings of Workshop on Constraint-Based Reasoning, Florida AI Research Symposium, Melbourne Beach, Florida, 1995.

E. Hyvönen, Evaluation of Cascaded Interval Function Constraints. Proceedings of Workshop on Constraint-Based Reasoning, Florida AI Research Symposium, Melbourne Beach, Florida, 1995.

E. Hyvönen , De Pascale, S., Shared Computations for Efficient Interval Function Evaluation. In: G. Alefeld, E. Frommer (eds), Scientific Computing, Computer Arithmetic and Validated Numerics, Akademie-Verlag, Berlin, Germany, 1996.

International journals

E. Hyvönen, De Pascale, S., Interval Constraint Spreadsheets — An Implementation for MS Excel. International Journal of Reliable Computing, supplement, 1995.

J. Pesonen J., E. Hyvönen, Interval Approach Challenges Monte Carlo Simulation. International Journal of Reliable Computing, 1996.

Other articles

Hyvönen, S. De Pascale, Next Generation Spreadsheet Computing: Range Solver for MS Excel. PC Artificial Intelligence.

Janne Pesonen, Risto Lehtinen, Monte Carlo Simulation, New Challenges in Firm-Wide Risk Management, Finance Line, A Financial Information Systems Newsletter, No. 2, 1997.

Technical documents

E. Hyvönen, S. De Pascale, Range Solver for MS Excel, User’s Guide, (VTT, Technical Research Centre of Finland, Information Technology, Espoo, Finland, 1996).

E. Hyvönen, S. De Pascale, ICE InC++, A Library for Interval Constraint Equations, Version Beta 1.1, (VTT, Technical Research Centre of Finland, Information Technology, Espoo, Finland, 1996).

M.Sc. Thesis

Stefano De Pascale, An implementation of interval computations on the spreadsheet, MSc. thesis/HY/Tietojenkäsittelytieteen laitos, 10.10.1996.

1.2 Industrial applications of genetic algorithms

Abstract

Results

Project information

Participants

• University of Vaasa
• KONE Oy
• ABB
• Kemira Safety
• Wärtsilä NSD

Project dates

Project volume

Project manager

Publications

Jarmo T. Alander, Jari Ylinen, Tapio Tyni (1995). Optimizing elevator group cont­rol parameters using distributed genetic algorithms. In D.W. Pearson, N.C. Steele, R.F. Albrecht (editors), Artificial Neural Nets and Genetic Algorithms, Alés (France), 19-21 April 1995, Pages 400-403 Springer-Verlag, Wien.

Jarmo T. Alander and Jouni Lampinen (1995). Shape optimization of diesel engine camshaft by genetic algorithm. In proceedings of MENDEL '95, First International Conference on Genetic Algorithms on the occasion of 130th anniversary of Mendel's laws in Brno, September 26-28, 1995, Brno, Czech Republic (1995), pages 5-10. Technical University of Brno, Faculty of Mechanical Engineering, Institute of Computer Science and Foundation Advanced Information Technology, Brno (Czech Republic). ISBN 80-214-0672-0.

Jarmo T. Alander, Mikael Nordman and Henri Setälä (1995). Register-level Hardware Design and Simulation of a Genetic Algorithm using VHDL.   In proceedings of MENDEL '95, First International Conference on Genetic Algorithms on the occasion of 130th anniversary of Mendel's laws in Brno, September 26-28, 1995, Brno, Czech Republic (1995), Pages 11-14. Technical University of Brno, Faculty of Mechanical Engineering, Institute of Computer Science and Foundation Advanced Information Technology, Brno (Czech Republic). ISBN 80-214-0672-0.

Jarmo T. Alander and Jouni Lampinen (1996). Improving   design of cam shape used in valvet­rain of internal-combustion engine using a genetic algorithm. In proceedings of MENDEL '96, 2nd International Mendel Conference on Genetic Algorithms, June 26-28, 1996, Brno, Czech Republic, pages 5-10. Technical University of Brno, Faculty of Mechanical Engineering, Institute of Computer Science and Foundation Anvanced Information Technology, Brno (Czech Republic).

Jarmo T. Alander and Jouni Lampinen (1996). Shape design and optimization of a diesel fuel injection cam by genetic algorithm. In Alander, Jarmo T. (ed.) (1996). Proceedings of the second Nordic Workshop on Genetic Algorithms and their Applications (2NWGA), 19-25 August 1996, Vaasa (Finland). Vaasan Yliopiston julkaisuja, selvityksiä ja raport­teja 11, pp. 195-204.

Jarmo T. Alander and Jouni Lampinen (1996). Shape optimization of internal-combusti­on engine valvecam by genetic algorithm. (Invited paper) In Proceedings of the 7th International DAAAM Symposium - product & manufacturing: flexibility, integration, intelligence (1996), 17-19 October 1996, Vienna (Austria), Pages 5-6, DAAAM International Vienna and Technical University of Vienna.

Jarmo T. Alander, Ghodrat Moghadampour and Jari Ylinen (1996). Comparison of eleva­tor allocation methods. In Alander, Jarmo T. (ed.) (1996). Proceedings of the second Nordic Workshop on Genetic Algorithms and their Applications (2NWGA), 19-25 August 1996, Vaasa (Finland). Vaasan Yliopiston julkaisuja, selvityksiä ja raportteja 11, pp. 211-214.

Jarmo T. Alander, Ghodrat Moghadampour and Jari Ylinen (1996). Solving the second order equation using genetic programming. In Alander, Jarmo T. (ed.) (1996). Proceedings of the second Nordic Workshop on Genetic Algorithms and their Applications (2NWGA), 19-25 August 1996, Vaasa (Finland). Vaasan Yliopiston julkai­suja, selvityksiä ja raportteja 11, pp. 215-218.

Pasi Törmänen (1996). Adaptive fuzzy PI-controller optimization by GA: A simulation study In Alander, Jarmo T. (ed.) (1996). Proceedings of the second Nordic Workshop on Genetic Algorithms and their Applications (2NWGA), 19-25 August 1996, Vaasa (Finland). Vaasan Yliopiston julkaisuja, selvityksiä ja raportteja 11, pp. 287-290.

Ghodrat Moghadampour (1996). Hissien ohjausjärjestelmien tilastollinen vertailu. Pro-gradu tutkielma. Vaasan Yliopisto, Tuotantotalouden ja tietotekniiman laitos, Vaasa.

Pasi Törmänen (1997). PID-säädön optimointi geneettisellä algoritmilla. Thesis for M.Sc. (Econ.) degree. University of Vaasa, Department of Information Technology. Vaasa (Finland). (In Finnish.)

Jarmo T. Alander, Juha Herajärvi, Ghodrat Moghadampour, Tapio Tyni and Jari Ylinen (1997). Genetic algorithms in elevator allocation problems. In Proceedings of ICANNCA '97, third international conference on artificial neural networks and genetic algorithms, 1-4 April 1997 (to appear), University of East Anglia, Norwich (UK).

Lampinen, Jouni (1997). Dieselmoottorin nokka-akseleiden mitoi­tusme­netelmän kehit­täminen (Development of design method for diesel engine cam­shafts). Thesis for M.Sc. (Econ.) degree. University of Vaasa, Department of Information Technology, Vaasa (Finland). (In Finnish.)

Lampinen, J. (1997). Choosing a shape representa­tion method for optimization of 2D shapes by genetic al­go­rithm. In Proceedings of the 3nd Nordic Workshop on Genetic Algorithms and their Applications, Alander, J.T. (ed.), pp. 305–319, Helsinki (Finland), Aug. 1997, Finnish Artificial Intelligence Society, available via anonymous ftp site ft­p.uwasa.fi file cs/3NWGA/Lampinen.ps.Z

Jarmo T. Alander and Jouni Lampinen (1997). A distributed implementation of genetic algorithm for cam shape optimization. In Barry H.V. Topping (ed.) (1997). Advances in computational mechanics with parallel and distributed processing (Proceedings of the first Euro Conference on Parallel and Distributed Computing for Computational Mechanics Lochinver, Highlands of Scotland 26th April - 1st May 1997), pages 209-217. Civil Comp Press, Edinburgh (Scotland). ISBN 0-948749-47-4.

Jarmo T. Alander and Jouni Lampinen (1997). On implementing CAD-systems based on existing simulators and optimization by genetic algorithm. In Ošmera, P. (ed.), Proceedings of MENDEL '97, 3rd International Mendel Conference on Genetic Algorithms, Optimization problems, Fuzzy logic, Neural Networks and Rough Sets, June 25-27, 1997, Brno, Czech Republic, pages 7-12. Technical University of Brno, Faculty of Mechanical Engineering, Institute of Automation and Computer Science, Brno (Czech Republic). ISBN 80-214-0884-7.

Alander, J.T. and Lampinen, J. (1997) Optimization of internal-combus­tion en­gine valve­cam by genetic al­go­rithm, In 8th International DAAAM Symposium - Intelligent Manufacturing & Automation, Katalinic, B. (ed.), pp. 5–6, Dubrovnik (Croatia), October 1997, DAAAM-International Vienna, Vienna (Austria), ISBN 3-901509-04-6.

Alander, J.T. and Lampinen, J. (1997). Cam shape optimization by genetic algorithm, In: Genetic Algorithms and Evolution Strategies in Engineering and Computer Science, Quagliarella, D. – Périaux, J. – Poloni, C. – Winter, G. (editors), pp. 153–174, John Wiley & Sons, Chichester (England), ISBN 0-471-97710-1.

Jouni Lampinen and Jarmo T. Alander (to appear). Shape design and optimization by genetic algorithm. In Barry H.V. Topping (ed.) (to appear). Proceedings of the second Euro-Conference on Parallel and Distributed Computing for Computational Mechanics, Sintra, Portugal, 4-9 April 1997.

Jouni Lampinen and Jarmo T. Alander (to be submitted). 2D Boundary Shape Optimization by Genetic Algorithm.

1.3 Probabilistic hybrid systems

Abstract

Results

The developed methods have been validated empirically by using both publicly available datasets that can be downloaded from the Internet, and proprietary data provided by the industrial partners. The empirical results demonstrate that for a wide range of different prediction tasks, the Bayesian approach outperforms the alternative prediction techniques, such as neural networks and decision trees. Similarly, the empirical optimization results suggest that the version of simulated annealing with automated parameter tuning produces better results than simulated annealing with manual parameter selection, or other sophisticated optimization methods, such as genetic algorithms or TABU search.

In software development, the main goal of the group has been the development of general prototype systems that can be used for demonstrating the benefits of the Bayesian approach. Two such demontration packages are currently available from the group's WWW home page: BAYDA, a tool for Bayesian discriminant analysis with automated feature selection, and D-SIDE, a prototype for a Bayesian decision support system. Both programs are provided with a JAVA user interface, which offers use of the software independent of the computer platform. The optimization methods developed by the group are currently been integrated into a fielded telecommunications software package   developed by Nokia.

Project information

Participants

• University of Helsinki
• Kone Elevators
• ABB Industry
• Nokia Telecommunications
• Nokia Research Center

Project dates

Project volume

Project manager

Publications

The home page of the CoSCo group is at http://www.cs.Helsinki.FI/research/cosco/

During the HYPE project, the CoSCo group produced 2 Ph.D. dissertations, 30 publications on international scientific forums, and a technology survey report on Bayesian networks, published by TEKES.

P.Grünwald, P.Kontkanen, P.Myllymäki, T.Silander, and H.Tirri, Minimum Encoding Approaches for Predictive Modeling. To appear in Proceedings of the 14th International Conference on Uncertainty in Artificial Intelligence (UAI'98), Madison, WI, USA, July 1998.

E.Koskimäki, J.Göös, P.Kontkanen, P.Myllymäki, and H.Tirri, Comparing Soft Computing Methods in Prediction of Manufacturing data. To appear in Proceedings of the 11th International Conference on Industrial and Engineering Applications of Artificial Intelligence & Expert Systems (IEA-98-AIE), Castellon, Spain, June 1998.

Tirri, What the heritage of Thomas Bayes has to offer for modern educational research? To appear in the next volume of Career Education Books, 1998.

Silander and H. Tirri, Bayesian classification. To appear in the next volume of Career Education Books, 1998.

Nokelainen, P. Ruohotie and H. Tirri, Professional Growth Determinants-Comparing Bayesian and linear approaches to classification. To appear in the next volume of Career Education Books, 1998.

Tirri and T.Silander, Stochastic complexity based estimation of missing elements in questionnaire data. Presented at the Annual American Educational Research Association Meeting (AERA'98), SIG Educational Statisticians, San Diego, 1998.

P.Myllymäki and H.Tirri, Prospects of Bayesian networks (in Finnish). Technology Report 58/98. Technology Development Center (TEKES), 1998.   

P.Kontkanen, P. Myllymäki, T. Silander, and H.Tirri, Bayes Optimal Instance-Based Learning. Pp. 77-88 in Machine Learning: ECML-98, Proceedings of the 10th European Conference, edited by C.Nédellec and C.Rouveirol. Vol. 1398 in Lecture Notes in Artificial Intelligence, Springer-Verlag, 1998.   

P.Kontkanen, P. Myllymäki, T. Silander, H.Tirri, and P. Grünwald, Bayesian and Information-Theoretic Priors for Bayesian Network Parameters. Pp. 89-94 in Machine Learning: ECML-98, Proceedings of the 10th European Conference, edited by C.Nédellec and C.Rouveirol. Vol. 1398 in Lecture Notes in Artificial Intelligence, Springer-Verlag, 1998.   

P.Kontkanen, P. Myllymäki, T. Silander, and H.Tirri, Batch Classifications with Discrete Finite Mixtures. Pp. 208-213 in Machine Learning: ECML-98, Proceedings of the 10th European Conference, edited by C.Nédellec and C.Rouveirol. Vol. 1398 in Lecture Notes in Artificial Intelligence, Springer-Verlag, 1998. style="mso-spacerun: yes">  

P.Kontkanen, P. Myllymäki, T. Silander, and H.Tirri, Bayesian Classification and Feature Selection with BAYDA. In ECML-98: Demonstration and poster papers, edited by C.Nédellec and C.Rouveirol. CSR-98-07, Technische Universität Chemnitz, 1998.   

P.Kontkanen, P. Myllymäki, T. Silander, H.Tirri, and P. Grünwald, Bayesian and Information-Theoretic Predictive Distributions for Bayesian Networks. Pp. 59-68 in Proceedings of the Seventh Belgian-Dutch Conference on Machine Learning (BeNeLearn '97), edited by W. Daelemans, P. Flach and A. van den Bosch. Tilburg, the Netherlands, October 1997.   

P.Kontkanen, P.Myllymäki, T.Silander, and H.Tirri, A Bayesian Approach for Retrieving Relevant Cases. Pp. 67-72 in Artificial Intelligence Applications (Proceedings of the EXPERSYS-97 Conference), edited by P.Smith. IITT International, Gournay sur Marne, 1997.   

Tirri, T.Silander and K.Tirri, Using neural networks for descriptive statistical analysis of educational data. Presented at the Annual American Educational Research Association Meeting (AERA'97), SIG Educational Statisticians, Chicago, 1997.   

Tirri, T.Silander and K.Tirri, Bayesian Finite Mixtures for nonlinear modeling of educational data. Presented at the Annual American Educational Research Association Meeting (AERA'97), Division D, Chicago, 1997.   

H.Tirri, Plausible Prediction by Bayesian Inference. Ph.D. Dissertation, Report A-1997-1, Department of Computer Science, University of Helsinki, December 1995. style="mso-spacerun: yes">  

P.Kontkanen, P.Myllymäki, and H.Tirri, Experimenting with the Cheeseman-Stutz Evidence Approximation for Predictive Modeling and Data Mining. Pp. 204-211 in Proceedings the Tenth International FLAIRS Conference (Daytona Beach, Florida, May 1997).   

P.Kontkanen, P.Myllymäki, T.Silander, and H.Tirri, A Bayesian Approach to Discretization. Pp. 265-268 in Proceedings of the European Symposium on Intelligent Techniques (Bari, Italy, March 1997).   

P.Kontkanen, P.Myllymäki, T.Silander, and H.Tirri, On the Accuracy of Stochastic Complexity Approximations. To appear in Causal Models and Intelligent Data Analysis, edited by A.Gammerman. Also: Pp. 103-117 in Proceedings of the Causal Models and Statistical Learning Seminar (London, UK, March 1997).   

P.Kontkanen, P. Myllymäki, T.Silander, and H.Tirri, Comparing Stochastic Complexity Minimization Algorithms in Estimating Missing Data. Pp. 81-90 in Proceedings of WUPES'97, the 4th Workshop on Uncertainty Processing (Prague, Czech Republic, January 1997).   

P.Kontkanen, P.Myllymäki, T.Silander, H.Tirri, and P.Grünwald, Comparing Predictive Inference Methods for Discrete Domains. Pp. 311-318 in Proceedings of the Sixth International Workshop on Artificial Intelligence and Statistics (Ft. Lauderdale, USA, January 1997).   

H.Tirri, P.Kontkanen, and P. Myllymäki, A Bayesian Framework for Case-Based Reasoning. Pp. 413-427 in Advances in Case-Based Reasoning (Proceedings of the 3rd European Workshop), edited by I.Smith and B.Faltings. Lecture Notes in Artificial Intelligence, Volume 1168, Springer-Verlag, Berlin Heidelberg, 1996.   

J.Lahtinen, P.Myllymäki, T.Silander, and H.Tirri, Empirical comparison of stochastic algorithms in a graph optimization problem. Pp. 45-59 in Proceedings of the Second Nordic Workshop on Genetic Algorithms and their Applications (Vaasa, Finland, August 1996), edited by J.Alander. University of Vaasa and the Finnish Artificial Intelligence Society, Vaasa, 1996.   

P.Kontkanen, P.Myllymäki and H.Tirri, Predictive Data Mining with Finite Mixtures. Pp. 176-182 in Proceedings of The Second International Conference on Knowledge Discovery and Data Mining (Portland, OR, August 1996).   

P.Kontkanen, P.Myllymäki and H.Tirri, Comparing Bayesian model class selection criteria by discrete finite mixtures. Pp. 364-374 in Information, Statistics and Induction in Science (Proceedings of the ISIS'96 Conference in Melbourne, Australia, August 1996), edited by D.L.Dowe, K.B.Korb, and J.J.Oliver. World Scientific, Singapore 1996.   

P.Kontkanen, P.Myllymäki and H.Tirri, Constructing Bayesian finite mixture models by the EM algorithm. NeuroCOLT Technical Report NC-TR-97-003.   

H.Tirri, P.Kontkanen and P.Myllymäki, Probabilistic Instance-Based Learning. Pp. 507-515 in Machine Learning: Proceedings of the Thirteenth International Conference, edited by L. Saitta. Morgan Kaufmann Publishers, San Francisco, CA, 1996.   

P.Kontkanen, P.Myllymäki and H.Tirri, Some experimental results with finite mixture models. Pp. 112-115 in Proceedings of the First European Conference on Highly Structured Stochastic Systems (Rebild, Denmark, May 1996).   

Myllymäki, Mapping Bayesian Networks to Stochastic Neural Networks: A Foundation for Hybrid Bayesian-Neural Systems. Ph.D. Dissertation, Report A-1995-1, Department of Computer Science, University of Helsinki, December 1995.   

H.Tirri and S.Mallenius, Optimizing the Hard Address Distribution for Sparse Distributed Memories. Pp. 1966-1970 in Proceedings of the IEEE International Conference on Neural Networks (Perth, November 1995).   

X.M.Song, H.Tirri, O.Aaltonen and A.Hase, A Discrete Radial Basis Function Network for Empirical Modeling of Soil Extraction Process. Pp. 17-20 in Proceedings of the International Conference on Engineering Applications of Neural Networks (EANN'95), 1995.   

Myllymäki, Mapping Bayesian Networks to Boltzmann Machines . Pp. 269-280 in Proceedings of Applied Decision Technologies 1995 (London, April 1995).   

Tirri, Replacing the Pattern Matching of an Expert System with a Neural Network. Pp. 47-62 in Intelligent Hybrid Systems, edited by S. Goonatilake and S. Khebbal. John Wiley & Sons, Chichester, 1995.   

Myllymäki and H. Tirri, Constructing computationally efficient Bayesian models via unsupervised clustering. Pp. 237-248 in Probabilistic Reasoning and Bayesian Belief Networks, edited by A.Gammerman. Alfred Waller Publishers, Suffolk, 1995.

1.4 Sales support system for synchronous machines

Results

A configuration application, AIMO (Artificial Intelligence for Machine Optimization), which utilizes expert knowledge, calculation programs and optimization was developed. The system automatically configures the synchronous machine based on the customer’s parameters. A genetic algorithm is used for optimization and fuzzy logic for evaluation of optimized technical parameters. Promising results concerning the pricing of the machines were also obtained. The pricing system is a collection of crisp or fuzzified expert rules and neural network models of factory data. The average errors between predicted and actual price were below five per cent.

Project information

Participants

• Automation laboratory in Helsinki University of Technology
• ABB \ Industry \ Machines

Project dates

Project volume

Project manager

Publications

Jouni Jaakkola: Development of a Synchronous Machines Sales Support Program, Master’s thesis, 1994.

E.Koskimäki and J.Göös. Electric machine dimensioning by global optimization. In Proceedings of the First International Conference on Knowledge Based Intelligent Electronic Systems, pages 308-312, Adelaide, Australia, 1997.

J.Göös and E.Koskimäki. Utilization of Artificial Intelligence Techniques in Making Offers. Presented in the Second Nordic Workshop on Genetic Algorithms and their Applications, 1996.

E.Koskimäki and J.Göös. Electric Machine Design by Fuzzy Fitness. The second Nordic Workshop on Genetic Algorithms and their Applications, 1996.

Esa Koskimäki et al. Comparing Soft Computing Methods in Prediction of Manufacturing Data, To be published in summer 1998.

Esa Koskimäki. A Genetic Algorithm for Capacity Analysis and Scheduling of Electric Machine Manufacturing, To be published in autumn 1998.

jukka.iivarinen@hut.fi
http://www.cis.hut.fi/neuronet/Tekes/1.shtml
Tuesday, 28-Nov-2000 15:28:05 EET