Analysis of source sparsity and recoverability for SCA based blind source separation

Yuanqing Li, Andrzej Cichocki, Shun Ichi Amari, Cuntai Guan

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

Abstract

One (of) important application of sparse component analysis (SCA) is in underdetermined blind source separation (BSS). Within a probability framework, this paper focuses on recoverability problem of underdetermined BSS based on a two-stage SCA approach. We consider a general case in which both sources and mixing matrix are randomly drawn. First, we present a recoverability probability estimate under the condition that the nonzero entry number of a source column vector is fixed. Next, we define the sparsity degree of a signal, and establish the relationship between the sparsity degree of sources and recoverability probability. Finally, we explain how to use the relationship to guarantee the performance of BSS. Several simulation results have demonstrated the validity of the probability estimation approach.

Original languageEnglish
Title of host publicationIndependent Component Analysis and Blind Signal Separation - 6th International Conference, ICA 2006, Proceedings
Pages831-837
Number of pages7
DOIs
Publication statusPublished - 2006
Externally publishedYes
Event6th International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2006 - Charleston, SC, United States
Duration: 5 Mar 20068 Mar 2006

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume3889 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference6th International Conference on Independent Component Analysis and Blind Signal Separation, ICA 2006
Country/TerritoryUnited States
CityCharleston, SC
Period5/03/068/03/06

Fingerprint

Dive into the research topics of 'Analysis of source sparsity and recoverability for SCA based blind source separation'. Together they form a unique fingerprint.

Cite this