Chapter

Advances in Neural Networks – ISNN 2005

Volume 3497 of the series Lecture Notes in Computer Science pp 466-471

Sequential Extraction Algorithm for BSS Without Error Accumulation

  • Qiang LiuAffiliated withLaboratory of Nonlinear Science, Institute of Mathematics, Fudan University
  • , Tianping ChenAffiliated withLaboratory of Nonlinear Science, Institute of Mathematics, Fudan University

* Final gross prices may vary according to local VAT.

Get Access

Abstract

Blind source separation (BSS) is an emerging research field in both theory and applications. In this paper we propose a kurtosis maximization algorithm–Sequential Extraction Algorithm, which can extract the source signals sequentially. This approach is based on an algorithm for separating one signal (Algorithm 1) and some technique to eliminate the accumulating errors, which often occur in the sequential extraction steps. In Algorithm 1, a new criterion to judge whether the separated signal is an original source signal, is proposed. In Sequential Extraction Algorithm, we propose a new approach to eliminate accumulating errors, which is caused in the sequential extraction process. This approach is based on the cost function involved in this algorithm, and thus, is different from those available in literature.