Connectionists: NMF and NTF alternative tools for Blind Source Separation and Multidimensional Decompositions/Factorizations

[A. Cichocki]

Re: NMF and  NTF alternative tools for Blind Source Separation
and/or  Multidimensional Decompositions/Factorizations

Dear  List  members,

We would like to bring to your attention the availability 
 on our web pages new upgraded versions of MATLAB toolboxes:

NTFLAB for  Signal Processing 

and 

NMFLAB for signal and image processing

http://www.bsp.brain.riken.jp/ICALAB/nmflab.html

http://www.bsp.brain.riken.jp/ICALAB/Guidebook-NMFLAB_SP.pdf

and associated list of publications and unpublished reports.


Nonnegative Matrix Factorization (NMF) and 
3D Nonnegative Tensor Factorization (NTF) can be considered  as
alternative  tools for some constrained BSS and  multi-dimensional 
signal/image/patterns decompositions problems  
of nonnegative components or factors  which are not necessary independent or sparse.

We proposed several novel algorithms and models, including 
multiplicative, fixed-point regularized ALS, projected gradient, 
exponentiated gradient algorithms.

 Especially, multi-layer (recurrent) NMF/ NTF algorithms 
with multi-start, multiple initialization methods provide 
improved reliability and excellent 
 performance for wide class of BSS  or 
Multi-dimensional Factorization applications.

Although, we have performed  successfully 
thousands of experiments and simulations for various benchmarks 
and real problems still could be some bugs or errors.
We would be  grateful for any critical comments 
or suggestions for improvements.


The references to our three MATLAB Toolboxes are given below:

[1] A. Cichocki and R. Zdunek, NMFLAB - 
MATLAB Toolbox for Non-Negative Matrix Factorization,
http://www.bsp.brain.riken.jp/ICALAB/nmflab.html

[2] A. Cichocki, R. Zdunek, and S. Amari, "Csiszar's Divergences for Non-Negative Matrix Factorization:
Family of New Algorithms", 6th International Conference on Independent Component Analysis and Blind
Signal Separation, Charleston SC, USA, March 5-8, 2006 Springer LNCS 3889, pp. 32-39. [.pdf]

[3] A. Cichocki, S. Amari, R. Zdunek, R. Kompass , G. Hori and Z. He: "Extended SMART Algorithms for
Non-Negative Matrix Factorization", 8th International Conference on Artificial 
Intelligence and Soft Computing,ICAISC, Zakopane, Poland, 25-29 June 2006, 
Springer Lectures Notes in Artificial Intelligence, Vol. 4029, pp.548-562. [.pdf]

[4] R. Zdunek, and A. Cichocki, "Non-Negative Matrix Factorization with Quasi-Newton Optimization",
8th International Conference on Artificial Intelligence and Soft Computing, ICAISC, Zakopane, Poland,
25-29 June, 2006,  Springer Lectures Notes in Artificial Intelligence, Vol. 4029, pp.870-879 [.pdf]

[5] A. Cichocki, R. Zdunek, and S. Amari, "New Algorithms for Non-Negative Matrix Factorization in
Applications to Blind Source Separation", 2006 IEEE International Conference on Acoustics, Speech, and
Signal Processing, ICASSP-2006, May 14-19, 2006, Toulouse, France, pp V-621-625 [.pdf]

[6] R. Zdunek and A. Cichocki, "Nonnegative Matrix Factorization with Constrained 
Second Order Optimization, (submitted for publication).

[7] A. Cichocki and R. Zdunek, "Nonnegative Matrix Factorization Using 
Projected Gradient Approaches", (ICONIP-2006, available on request).

[8] A. Cichocki and R. Zdunek, "Multilayer Nonnegative Matrix Factorization", 
Electronics Letters, Vol. 42, No. 16 (2006), pp. 947-948.

[9] A. Cichocki, R. Zdunek, R. Plemmons and S. Amari, 
"Novel Multi-layer Nonnegative 3D Tensor
Factorization with Sparsity Constraints" (submitted).

[10] A. Cichocki, "Generalized Component Analysis and Blind Source Separation Methods for
Analyzing Multichannel Brain Signals" - preliminary version of a chapter in the monograph,
"Statistical and Process Models of Cognitive Aging", Notre Dame Series
on Quantitative Methods, Editors: M.J.  Wenger and C. Schuster, (Mahwah, NJ: Erlbaum, 2007) [http <http://www.bsp.brain.riken.jp/publications/2006/CichockiNDSQM-2006.pdf>]


Andrzej Cichocki
==========
Laboratory for Advanced Brain Signal Processing
Riken, Brain Science Institute, JAPAN
Wako Shi, Saitama 351-0198
==================
http://www.bsp.brain.riken.jp/index.php
http://www.bsp.brain.riken.jp/~cia/