Blind source separation based on collective neurodynamic optimization approach

Abstract

In this paper, a collective neurodynamic optimization approach is proposed to blind source separation via tensor decomposition. Tensor decompositions have a lot of success in many fields, such as blind source separation, remote sensing image processing, text mining, linear regression, and feature extraction. However, decomposition process would cost much time and usually trap into the local minima. To solve this problem, a novel collective neurodynamic optimization (CNO) approach is presented by adopting a group of recurrent neural networks (RNN) in framework of particle swarm optimization (PSO). Through iteratively improving the best position of each RNN, the global optimal solutions of tensor decomposition are found. Blind source separation experiments confirm the validity and higher performance of the proposed algorithm in comparison to the state-of-the-art algorithms.