PSO-based infomax algorithm for frequency-domain blind source separation

This paper discusses an improved infomax algorithm for independent component analysis to process complex-valued signals in frequency-domain blind source separation (BSS). It is known that the choice of the step-size affects the speed of convergence in infomax algorithm. In this paper, a step-size adjustment method is proposed for infomax algorithm, which is based on particle swarm optimization (PSO) technique. Our discussion is supported by experimental results for separating speech signals, which show that the proposed approach exhibits rapid convergence than fixed step-size methods.