This is a MATLAB code that performs
Independent Component Analysis
using a deflation approach.

	Y = dICA(X, d, s)

Input Formats

    X = n x m dimensional input matrix [x1 x2 ... xm] 
	(observed signals) 
    where	m = number of data
             	xi = n x 1 dimensional column vector
             	n = number of sensors

    d = number of components to be extracted

    s =  0  for non-selective extraction   
    s = -1 for extraction of source signals with negative kurtosis
    s = +1 for extraction of source signals with positive kurtosis

Output Formats

   Y = d x m dimensional output matrix (extracted signals)

Reference Papers

[1] Ruck Thawonmas, Andrzej Cichocki, and Shun-ichi Amari, 
"A Cascade Neural Network for Blind Signal Extraction without
Spurious Equilibria," IEICE Trans. on Fundamentals of Electronics, 
Communications and Computer Sciences, 
vol. E81-A, no. 9, September, pp. 1833-1846, 1998. 

[2] Ruck Thawonmas, "A Neural Network Model for Projection 
Pursuit, " Workshop on Recent Advances in Soft Computing 1999,
July 1-2 1999, Leicester, UK, published in Advances in Soft 
Computing, R. John and R. Birkenhead, Eds., Springer-Verlag,
Heidelberg, pp. 28-33, 2000. 

All rights are reserved for Ruck Thawonmas, July 10, 2000,
Kochi University of Technology, Japan.
http://www.info.kochi-tech.ac.jp/ruck

-------------------------Have Fun!-------------------------