Abstract
The FastICA algorithm is a popular procedure for independent component analysis and blind source separation. Recently, several of its convergence properties have been elucidated, including its average convergence performance and its finite-sample behavior. In this paper, we examine the kurtosis-based algorithm version for two-source mixtures with equal-kurtosis sources, proving that the single-unit FastICA algorithm has dynamical behavior that is identical to the Newton-based Rayleigh Quotient Iteration for finding an eigenvector of a symmetric matrix. We also derive a bound on the average inter-channel interference indicating that the initial convergence rate of FastICA is linear with a rate of (1/3). A simulation indicates its convergence performance.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Hyvärinen, A., Oja, E.: A fast fixed-point algorithm for independent component analysis. Neural Computation 9(7), 1483–1492 (1997)
Hyvarinen, A., Karhunen, J., Oja, E.: Independent Component Analysis. Wiley, New York (2001)
Oja, E.: Convergence of the symmetrical FastICA algorithm. In: Proc. 9th Int. Conf. Neural Inform. Processing, Singapore, vol. 3, pp. 1368–1372 (November 2002)
Douglas, S.C.: On the convergence behavior of the FastICA algorithm. In: Proc. Fourth Symp. Indep. Compon. Anal. Blind Signal Separation, Kyoto, Japan, pp. 409–414 (April 2003)
Douglas, S.C.: A statistical convergence analysis of the FastICA algorithm for twosource mixtures. In: Proc. 39th Asilomar Conf. Signals, Syst., Comput., Pacific Grove, CA (October 2005)
Douglas, S.C., Yuan, Z., Oja, E.: Average convergence behavior of the FastICA algorithm for blind source separation. In: Proc. 6th Int. Conf. Indep. Compon. Anal. Blind Source Separation, Charleston, SC (March 2006) (to appear)
Koldovsky, Z., Tichavsky, P., Oja, E.: Cramer-Rao lower bound for linear independent component analysis. In: Proc. IEEE Int. Conf. Acoust., Speech, Signal Processing, Philadelphia, PA, vol. 3, pp. 581–584 (March 2005)
Edelman, A., Arias, T.A., Smith, S.T.: The geometry of algorithms with orthogonality constraints. SIAM J. Matrix Anal., Appl. 20, 303–353 (1999)
Golub, G.H., Van Loan, C.F.: Matrix Computations, 3rd edn. Johns Hopkins, Baltimore (1996)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2006 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Douglas, S.C. (2006). Relationships Between the FastICA Algorithm and the Rayleigh Quotient Iteration. In: Rosca, J., Erdogmus, D., Príncipe, J.C., Haykin, S. (eds) Independent Component Analysis and Blind Signal Separation. ICA 2006. Lecture Notes in Computer Science, vol 3889. Springer, Berlin, Heidelberg. https://doi.org/10.1007/11679363_97
Download citation
DOI: https://doi.org/10.1007/11679363_97
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-32630-4
Online ISBN: 978-3-540-32631-1
eBook Packages: Computer ScienceComputer Science (R0)