Independent Component Analysis Using Convex Divergence

The convex divergence is used as a surrogate function for obtaining a class of ICA algorithms (Independent Component Analysis) called the f-ICA. The convex divergence is a super class of α-divergence, which is a further upper family of Kullback-Leibler divergence or mutual information. Therefore, the f-ICA contains the α-ICA and the minimum mutual information ICA. In addition to theoretical interest of generalization, the f-ICA contains a subset faster than the minimum mutual information ICA. It is found that this speed control is equivalent to the α-ICA. Finally, applications to brain fMRI map’s distillation is presented.

[1]  J. Cardoso,et al.  Blind beamforming for non-gaussian signals , 1993 .

[2]  Terrence J. Sejnowski,et al.  Blind separation and blind deconvolution: an information-theoretic approach , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[3]  Shun-ichi Amari,et al.  Adaptive Online Learning Algorithms for Blind Separation: Maximum Entropy and Minimum Mutual Information , 1997, Neural Computation.

[4]  Christian Jutten,et al.  Blind separation of sources, part I: An adaptive algorithm based on neuromimetic architecture , 1991, Signal Process..

[5]  Shun-ichi Amari,et al.  Natural Gradient Works Efficiently in Learning , 1998, Neural Computation.

[6]  Thomas M. Cover,et al.  Elements of Information Theory , 2005 .

[7]  Yasuo Matsuyama,et al.  Independent component analysis by convex divergence minimization: applications to brain fMRI analysis , 2001, IJCNN'01. International Joint Conference on Neural Networks. Proceedings (Cat. No.01CH37222).

[8]  Pierre Comon,et al.  Independent component analysis, A new concept? , 1994, Signal Process..

[9]  Aapo Hyvärinen,et al.  Fast and robust fixed-point algorithms for independent component analysis , 1999, IEEE Trans. Neural Networks.

[10]  S Makeig,et al.  Spatially independent activity patterns in functional MRI data during the stroop color-naming task. , 1998, Proceedings of the National Academy of Sciences of the United States of America.

[11]  Yasuo Matsuyama,et al.  /spl alpha/-EM algorithm and /spl alpha/-ICA learning based upon extended logarithmic information measures , 2000, Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium.

[12]  Ali Mansour,et al.  Blind Separation of Sources , 1999 .

[13]  Pierre Comon.  Independent component analysis - a new concept? signal processing , 1994 .