Blind separation of statistically independent signals with mixed sub-gaussian and super-gaussian probability distributions

Muhammad Tufail, Masahide Abe, Masayuki Kawamata

Research output: Contribution to journalConference articlepeer-review

Abstract

In the context of Independent Component Analysis (ICA), we propose a simple method for online estimation of activation functions in order to blindly separate instantaneous mixtures of sub-Gaussian and super-Gaussian signals. An adequate choice of these activation functions is necessary not only for a successful source separation (using relative gradient algorithm), but also to achieve sufficient level of cross-talk index. To accomplish this, we employ a simple parameterized model for the probability density functions of sources. The parameter of this distribution model (for each estimated source signal) is adapted online by minimizing the mutual information while the activation functions are obtained as the associated score functions. Furthermore, a modified relative gradient algorithm is derived that exhibits an isotropic convergence (near the desired solution) independent of the statistics of sources. Some simulation results are given to demonstrate the effectiveness of the presented methods.

Original languageEnglish
Article number1465265
Pages (from-to)3027-3030
Number of pages4
JournalProceedings - IEEE International Symposium on Circuits and Systems
DOIs
Publication statusPublished - 2005 Dec 1
EventIEEE International Symposium on Circuits and Systems 2005, ISCAS 2005 - Kobe, Japan
Duration: 2005 May 232005 May 26

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Blind separation of statistically independent signals with mixed sub-gaussian and super-gaussian probability distributions'. Together they form a unique fingerprint.

Cite this