Publications using denoising source separation

We collect here the work that has been published using DSS. If you publish something, please let us know. The central papers for the theory of DSS are:

Denoising source separation, J. Särelä and H. Valpola, 2005.

Accurate, fast and stable denoising source separation algorithms, H. Valpola and J. Särelä, 2004.

Fast algorithms for Bayesian independent component analysis, H. Valpola and P. Pajunen, 2000.

2006

Exploratory analysis of climate data using source separation methods.

A. Ilin, H. Valpola and E. Oja.

Neural Networks, 19(2):155-167, 2006.

[pdf 3.4 MB] [html]

This article combines the PKDD'05 and IJCNN'05 articles. Some new results are presented.

 

Separation of nonlinear image mixtures by denoising source separation.

M.S.C. Almeida, H. Valpola and J. Särelä.

In Proceedings of the 6th International Conference on Independent Component Analysis and Blind Source Separation, ICA 2006, Charleston, SC, USA, pp 8-16, 2006.

[pdf 362 kB]

The denoising source separation framework is extended to nonlinear separation of image mixtures. MLP networks are used to model the nonlinear unmixing mapping. Learning is guided by a denoising function which uses prior knowledge about the sparsity of the edges in images. The main benefit of the method is that it is simple and computationally efficient. Separation results on a real-world image mixture proved to be comparable to those achieved with MISEP.

 

2005

Frequency-Based Separation of Climate Signals.

A. Ilin and H. Valpola.

In the proceedings of the 9th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD 2005), Porto, Portugal, pp. 519-526, 2005.

 

Semiblind source separation of climate data detects El Niño as the component with the highest interannual variability.

A. Ilin, H. Valpola and E. Oja.

In Proceedings of the International Joint Conference on Neural Networks (IJCNN 2005), Montréal, Québec, Canada, pp. 1722-1727, 2005.

[pdf 1.4 MB]

DSS can find features with certain characteristics. It turns out that in a certain large climate dataset, the phenomenon with the highest interannual variability is the well-known El Niño. Many other intersting phenomena are found, too. The linear DSS method we used here can only find a signal subspace, not a rotation in it. Real separation results are published in the PKDD 2005 paper

 

Development of representations, categories and concepts-- a hypothesis.

H. Valpola.

In Proceedings of the 6th IEEE International Symposium on Computational Intelligence in Robotics and Automation, CIRA 2005, Helsinki, Finland, pp. 593-599, 2005

[pdf 72 kB]

This is Harri's brain-related "visions-and-ideas paper". He refers to simulation results in many of the of the machine-learning-oriented DSS papers, Deco's results with attention model and to some biological findings. Based on these, he proposes how the brain could learn concepts and representations in active interaction with the world.

 

Source localization of low- and high-amplitude alpha activity: A segmental and DSS analysis.

S. Borisov, A. Ilin, R. Vigário and A. Kaplan.

In the 11th Annual Meeting of Organization for Human Brain Mapping, Toronto, Canada, June, 12-16, 2005.

According to the results of the presented work, it is possible to assume that both low- and high-amplitude EEG alpha activity may have both common and distinctive bioelectrical sources in the brain. The different spatial patterns of sources found for low- and high-amplitude alpha activity may suggest that these populations of alpha-activity have their own nature and could perform different physiological functions.

 

Denoising source separation: a novel approach to ICA and feature extraction using denoising and Hebbian learning.

J. Särelä and H. Valpola.

In AI 2005, special session on correlation learning, pp.45-56.

[12 page paper, pdf 1.7 MB] [2 page abstact, pdf 800 kB] [slides, pdf 2.7 MB]

A comprehensive talk describing DSS in general and its biological relevance.

 

Single trial denoising source separation of event-related fields.

R. Vigário and J. Särelä.

In Tandem Workshop on Advanced Methods of Electrophysiological Signal Analysis (Part A) and Symbol Grounding? Dynamical Systems Approaches to Language (Part B), Potsdam, Germany, March 2005.

This paper applies DSS to the analysis of single trial event-related MEG signals. The denoising method has similarity to one in the cardiac subspace experiment in the DSS paper.

 

Denoising source separation.

J. Särelä and H. Valpola.

Journal of Machine Learning Research, 6:233-272, 2005.

[abstract] [pdf 2MB]

This is a comprehensive machine learning perspective to DSS.

 

2004

A denoising source separation based approach to interference cancellation for DS-CDMA array systems.

K. Raju and J. Särelä.

In Proceedings of the 38th Asilomar Conference on Signals, Systems, and Computers, Pacific grove, CA, USA, pp.~1111 -- 1114, 2004.

This paper applies DSS to DS-CDMA interference cancellation and channel estimation.

 

Accurate, fast and stable denoising source separation algorithms.

H. Valpola and J. Särelä.

In Proceedings of the 5th International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2004, Granada, Spain, pp. 65-72, 2004.

[abstract]

Even faster than the famous FastICA and robust, too. We have seen that this method gives nice separation results for the climate-phenomenon subspace found in this paper (but haven't published the results yet).

 

Denoising source separation: from temporal to contextual invariance.

H. Valpola and J. Särelä.

Presented in Early Cognitive Vision Workshop, Isle of Skye, Scotland, 2004.

[pdf 46 kB (abstract)] [pdf 2.6 MB (poster about DSS)] [pdf 89 kB (poster about context-guided denoising)]

The first poster gives an overview of DSS and the second explains how context can be used for denoising, promoting the development of invariant representations.

 

Behaviourally meaningful representations from normalisation and context-guided denoising.

H. Valpola.

AI Lab technical report, University of Zurich, 2004.

[abstract]

Invariant features resembling complex-cell properties are known to develop if temporal slowness is the learning criterion. Harri argues that this is a special case of expectation and shows that lateral expectation from adjacent image location will also produce complex-cell-like feature detectors. It also turned out that the expectation-driven learning with DSS resembles in many ways Deco's model for attention. Finding invariant features and attentional filtering are both selection processes, only on different timescales. Harri discusses the connections and proposes that normalisation of activations of competing neuron assemblies makes attentional process robust in the same way as decorrelation of inputs helps DSS.

 

Earlier work

A fast semi-blind source separation algorithm.

H. Valpola and J. Särelä.

In Publications in Computer and Information Science, Report A66, Helsinki University of Technology, Espoo, Finland, 4 p., 2002.

[pdf 140 kB]

Here we put the basic idea of DSS down before starting to write the JMLR article.

 

Dynamical factor analysis of rhythmic magnetoencephalographic activity.

J. Särelä, H. Valpola, R. Vigário and E. Oja.

In Proceedings of the Third International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2001, San Diego, CA USA, pp. 451 -- 456, 2001.

[pdf 560 kB]

This paper actually proposes a variational Bayesian method for separation of dynamic sources. The initialisation was done using DSS. It seems that DSS was doing the actual work in the separation...

 

Fast algorithms for Bayesian independent component analysis.

H. Valpola and P. Pajunen.

In Proceedings of the Second International Workshop on Independent Component Analysis and Blind Signal Separation, ICA 2000, Helsinki, Finland, pp. 233-237, 2000.

[html] [pdf 493 kB]

The first publication of the method that became DSS. Harri wanted to include this in his thesis and therefore used variational Bayesian methods for denoising.