Nonlinear blind source separation using a radial basis function network

This paper proposes a novel neural-network approach to blind source separat ion in nonlinear mixture, The approach utilizes a radial basis function (RB F) neural-network to approximate the inverse of the nonlinear mixing mappin g which is assumed to exist and able to be approximated using an RBF networ k. A contrast function which consists of the mutual information and partial moments of the outputs of the separation system, is defined to separate th e nonlinear mixture, The minimization of the contrast function results in t he independence of the outputs with desirable moments such that the origina l sources are separated properly. Two learning algorithms for the parametri c RBF network are developed by using the stochastic gradient descent method and an unsupervised clustering method. By virtue of the RBF neural network , this proposed approach takes advantage of high learning convergence rate of weights in the hidden layer and output layer, natural unsupervised learn ing characteristics, modular structure, and universal approximation capabil ity. Simulation results are presented to demonstrate the feasibility, robus tness, and computability of the proposed method.