How fast is FastICA?

The present contribution deals with the statistical tool of Independent Component Analysis (ICA). The focus is on the deflation approach, whereby the independent components are extracted one after another. The kurtosis-based FastICA is arguably one of the most widespread methods of this kind. However, its features, particularly its speed, have not been thoroughly evaluated or compared, so that its popularity seems somewhat unfounded. To substantiate this claim, a simple quite natural modification is put forward and assessed in this paper. It merely consists of performing exact line search optimization of the contrast function. Speed is objectively measured in terms of the computational complexity required to reach a given source extraction performance. Illustrative numerical results demonstrate the faster convergence and higher robustness to initialization of the proposed approach, which is thus referred to as RobustICA.

[1]  A. Ferreol,et al.  Comparative performance analysis of eight blind source separation methods on radiocommunications signals , 2004, 2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541).

[2]  Pierre Comon,et al.  Blind channel equalization with algebraic optimal step size , 2005, 2005 13th European Signal Processing Conference.

[3]  Pierre Comon,et al.  Improved contrast dedicated to blind separation in communications , 1997, 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[4]  Jitendra K. Tugnait,et al.  Identification and deconvolution of multichannel linear non-Gaussian processes using higher order statistics and inverse filter criteria , 1997, IEEE Trans. Signal Process..

[5]  Erkki Oja,et al.  Independent Component Analysis , 2001 .

[6]  P. Comon.  Independent Component Analysis , 1992 .

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

[8]  D. Godard,et al.  Self-Recovering Equalization and Carrier Tracking in Two-Dimensional Data Communication Systems , 1980, IEEE Trans. Commun..

[9]  P. Comon,et al.  Blind and semi-blind equalization based on the constant power criterion , 2005, IEEE Transactions on Signal Processing.

[10]  Aapo Hyvärinen,et al.  A Fast Fixed-Point Algorithm for Independent Component Analysis , 1997, Neural Computation.

[11]  Pierre Comon,et al.  Optimal Step-Size Constant Modulus Algorithm , 2008, IEEE Transactions on Communications.

[12]  Jean-Franois Cardoso.  High-Order Contrasts for Independent Component Analysis , 1999, Neural Computation.

[13]  Erkki Oja,et al.  Performance analysis of the FastICA algorithm and Crame/spl acute/r-rao bounds for linear independent component analysis , 2006, IEEE Transactions on Signal Processing.

[14]  Nathalie Delfosse,et al.  Adaptive blind separation of independent sources: A deflation approach , 1995, Signal Process..

[15]  P. Comon,et al.  Blind separation of discrete sources , 1998, IEEE Signal Processing Letters.

[16]  P. Comon,et al.  INDEPENDENT COMPONENT ANALYSIS , CONTRASTS , AND CONVOLUTIVE MIXTURES , 2003 .

[17]  E. Oja,et al.  Corrections to "Performance Analysis of the FastICA Algorithm and Cram´ er-Rao Bounds for Linear Independent Component Analysis", TSP 04/06 , 2007 .

[18]  F. A. Seiler,et al.  Numerical Recipes in C: The Art of Scientific Computing , 1989 .

[19]  Aapo Hyvärinen,et al.  A family of fixed-point algorithms for independent component analysis , 1997, ICASSP.

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

[21]  A. Hyvarinen.  A family of fixed-point algorithms for independent component analysis , 1997, 1997 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[22]  Jitendra K. Tugnait,et al.  Comments on 'New criteria for blind deconvolution of nonminimum phase systems (channels)' , 1992, IEEE Trans. Inf. Theory.

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