Neural network schemes for blind separation of sources from nonlinear mixtures

W. L. Woo, S. Sali

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Most of the existing algorithms are based on the ideal situation where the mixture is merely a linear transformation of the source signals and the demixer is simply a linear network, in this paper, nonlinear techniques are presented for instantaneous blind signal separation using information thearetic approach combined with (nonlinear) neural networks. Firstly, we addressed the issue of the modelling the mixture for both linear and nonlinear transformation of the source signals. Secondly, we derived the required algorithm to train the variable gradient multilayer perceptron (MLP) based on Lie group. In the past, most of the existing demixers employed a fixed gradient. Finally, computer simulations are being carried out to compare the performance of the linear and nonlinear demixer when the underlying mixture of the source signals is either linear or nonlinear.

Original languageEnglish
Title of host publication2002 14th International Conference on Digital Signal Processing Proceedings, DSP 2002
EditorsA.N. Skodras, A.G. Constantinides
PublisherIEEE
Pages1227-1234
Number of pages8
Volume2
ISBN (Print)0780375033
DOIs
Publication statusPublished - 7 Nov 2002
Event14th International Conference on Digital Signal Processing, DSP 2002 - Santorini, Hellas, Greece
Duration: 1 Jul 20023 Jul 2002

Conference

Conference14th International Conference on Digital Signal Processing, DSP 2002
CountryGreece
CitySantorini, Hellas
Period1/07/023/07/02

Fingerprint Dive into the research topics of 'Neural network schemes for blind separation of sources from nonlinear mixtures'. Together they form a unique fingerprint.

Cite this