Globally convergent Newton algorithms for blind decorrelation

Abstract

This paper presents novel Newton algorithms for the blind adaptive decorrelation of real and complex processes. They are globally convergent and exhibit an interesting relation ship with the natural gradient algorithm for blind decorre lation and the Goodall learning rule. Indeed, we show that these two later algorithms can be obtained from their New ton decorrelation versions when an exact matrix inversion is replaced by an iterative approximation to it.