ITE / code / shared / embedded / SWICA / README

SWICA (Schweizer-Wolff contrast for Independent Component Analysis) algorithm is implemented mostly in Matlab with some subroutines using Mex/C++.  Mex/C++ routines are also duplicated in Matlab code, but I STRONGLY suggest compiling C++ files and using the binaries instead.

List of files:
SWICA_solve.m -- envelop routine for running the software;

SW_contrast.m -- envelop routine for computing SW measure of dependence, the contrast used in SWICA;

SW_sigma.m    -- computation of SW sigma measure of dependence using empirical copulas;

SW_kappa.m    -- computation of SW kappa measure of dependence using empirical copulas;

SW_demo.m     -- a simple example of how to use SWICA

SWICA is distributed under BSD license.  See file LICENSE for details.

If you use the software, please use the following citation:
Sergey Kirshner, Barnabas Poczos, "ICA and ISA Using Schweizer-Wolff Measure of Dependence," Proceedings of the Twenty Fifth Conference on Machine Learning, 2008.

Sergey Kirshner (
May 7, 2008