Two-stage neural network for blind sources separation

Author

Choi, Seungjin ; Liu, Ruey- Wen

Author_Institution

Dept. of Electr. Eng., Notre Dame Univ., IN, USA

Volume

2

fYear

1996

Firstpage

827

Abstract

In this paper, an on-line implementation of the simultaneous diagonalization (SD) of two different symmetric matrices is addressed. A two-stage neural network which consists of self-normalizing decorrelation and extended Oja´s rule, is presented for an on-line implementation of SD. The SD of the 2nd- and 4th-order moment matrices is known as one solution to the blind sources separation problem. It will be shown that the two-stage network presented can recover the source signals from a linear mixture without the knowledge of the mixing matrix and the distribution of the source signals

Keywords

correlation methods; higher order statistics; matrix algebra; neural nets; signal processing; blind sources separation; extended Oja´s rule; fourth-order moment matrices; linear mixture; moment matrices; on-line implementation; second-order moment matrices; self-normalizing decorrelation; simultaneous diagonalization; symmetric matrices; two-stage neural network; Covariance matrix; Decorrelation; Higher order statistics; Intelligent networks; Laboratories; Neural networks; Signal analysis; Symmetric matrices; Vectors; Visual perception;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1996., IEEE 39th Midwest symposium on

Conference_Location

Ames, IA

Print_ISBN

0-7803-3636-4

Type

conf

DOI

10.1109/MWSCAS.1996.588042

Filename

588042