Comparison between Algorithms of Extraction based on the Nonstationarity of the Sources

J. Charneco Fernndez (Spain)

Keywords

Blind extraction, nonstationarity signals, high order statistics, adaptative algorithm.

Abstract

This article is based on the blind extraction of signals from the observations of a linear and instantaneous mixture. Criteria that use the non stationarity of the sources are analyzed. One use the algorithm of natural gradient to optimize one of these criteria and its result compares the one of the algorithm of fixed point. It is verified that the gradient algorithm allows the simultaneous extraction of the desired number of sources

Important Links:

• DOI:
• From Proceeding (413) Neural Networks and Computational Intelligence - 2004

Go Back