Title :
Infomax and maximum likelihood for blind source separation
Author :
Cardoso, Jean-François
Author_Institution :
Ecole Normale Superieure, Paris, France
fDate :
4/1/1997 12:00:00 AM
Abstract :
Algorithms for the blind separation of sources can be derived from several different principles. This article shows that the infomax (information-maximization) principle is equivalent to the maximum likelihood. The application of the infomax principle to source separation consists of maximizing an output entropy.
Keywords :
maximum entropy methods; maximum likelihood estimation; signal processing; blind source separation algorithms; infomax principle; information maximization; maximum likelihood; output entropy maximization; Blind source separation; Distribution functions; Entropy; H infinity control; Maximum likelihood estimation; Parametric statistics; Probability density function; Random variables; Source separation;
Journal_Title :
Signal Processing Letters, IEEE
