Novel dependence measure for dependent component analysis

Abstract

The purpose of this paper is to develop nonparametric blind signal separation (BSS) algorithm for linear dependent source signals, which is proposed under the framework of contrast method as in independent component analysis (ICA). The contrast function is derived from the Schweizer-Wolff measure of pairwise dependence between the variables. Simulation results show that the proposed algorithm is able to separate the dependent signals and yield ideal performance.