A novel Kurtosis-Dependent Parameterized Independent Component Analysis algorithm

Abstract

In the framework of natural gradient, a novel Kurtosis-Dependent Parameterized Independent Component Analysis (KDPICA) algorithm is proposed, which can separate the mixture of super- and sub-Gaussian sources. Two kinds of new probability density models are proposed, which can provide wider ranges especially for sub-Gaussian kurtosis. According to kurtosis value of source and whitening, model parameters are adaptively calculated which can be used to estimate super- and sub-Gaussian source distributions and its corresponding score functions directly. According to stability analysis, the ranges of model parameters are fixed which confirm KDPICA algorithm stable. The experiment shows the proposed algorithm has better performance than some proposed algorithms. © Springer-Verlag Berlin Heidelberg 2006.