Estimating functions for blind separation when sources have variance dependencies

Motoaki Kawanabe, Klaus Muller

Research output: Contribution to journalArticle

20 Citations (Scopus)

Abstract

A blind separation problem where the sources are not independent, but have variance dependencies is discussed. For this scenario Hyvärinen and Hurri (2004) proposed an algorithm which requires no assumption on distributions of sources and no parametric model of dependencies between components. In this paper, we extend the semiparametric approach of Amari and Cardoso (1997) to variance dependencies and study estimating functions for blind separation of such dependent sources. In particular, we show that many ICA algorithms are applicable to the variance-dependent model as well under mild conditions, although they should in principle not. Our results indicate that separation can be done based only on normalized sources which are adjusted to have stationary variances and is not affected by the dependent activity levels. We also study the asymptotic distribution of the quasi maximum likelihood method and the stability of the natural gradient learning in detail. Simulation results of artificial and realistic examples match well with our theoretical findings.

Original languageEnglish
JournalJournal of Machine Learning Research
Volume6
Publication statusPublished - 2005 Jul 21
Externally publishedYes

Fingerprint

Estimating Function
Blind source separation
Blind Source Separation
Dependent
Independent component analysis
Quasi-maximum Likelihood
Maximum likelihood
Maximum Likelihood Method
Parametric Model
Asymptotic distribution
Gradient
Scenarios
Simulation

Keywords

  • Blind source separation
  • Estimating functions
  • Independent component analysis
  • Semiparametric statistical models
  • Variance dependencies

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Estimating functions for blind separation when sources have variance dependencies. / Kawanabe, Motoaki; Muller, Klaus.

In: Journal of Machine Learning Research, Vol. 6, 21.07.2005.

Research output: Contribution to journalArticle

@article{50aafe91e49a429887045a6270af965c,
title = "Estimating functions for blind separation when sources have variance dependencies",
abstract = "A blind separation problem where the sources are not independent, but have variance dependencies is discussed. For this scenario Hyv{\"a}rinen and Hurri (2004) proposed an algorithm which requires no assumption on distributions of sources and no parametric model of dependencies between components. In this paper, we extend the semiparametric approach of Amari and Cardoso (1997) to variance dependencies and study estimating functions for blind separation of such dependent sources. In particular, we show that many ICA algorithms are applicable to the variance-dependent model as well under mild conditions, although they should in principle not. Our results indicate that separation can be done based only on normalized sources which are adjusted to have stationary variances and is not affected by the dependent activity levels. We also study the asymptotic distribution of the quasi maximum likelihood method and the stability of the natural gradient learning in detail. Simulation results of artificial and realistic examples match well with our theoretical findings.",
keywords = "Blind source separation, Estimating functions, Independent component analysis, Semiparametric statistical models, Variance dependencies",
author = "Motoaki Kawanabe and Klaus Muller",
year = "2005",
month = "7",
day = "21",
language = "English",
volume = "6",
journal = "Journal of Machine Learning Research",
issn = "1532-4435",
publisher = "Microtome Publishing",

}

TY - JOUR

T1 - Estimating functions for blind separation when sources have variance dependencies

AU - Kawanabe, Motoaki

AU - Muller, Klaus

PY - 2005/7/21

Y1 - 2005/7/21

N2 - A blind separation problem where the sources are not independent, but have variance dependencies is discussed. For this scenario Hyvärinen and Hurri (2004) proposed an algorithm which requires no assumption on distributions of sources and no parametric model of dependencies between components. In this paper, we extend the semiparametric approach of Amari and Cardoso (1997) to variance dependencies and study estimating functions for blind separation of such dependent sources. In particular, we show that many ICA algorithms are applicable to the variance-dependent model as well under mild conditions, although they should in principle not. Our results indicate that separation can be done based only on normalized sources which are adjusted to have stationary variances and is not affected by the dependent activity levels. We also study the asymptotic distribution of the quasi maximum likelihood method and the stability of the natural gradient learning in detail. Simulation results of artificial and realistic examples match well with our theoretical findings.

AB - A blind separation problem where the sources are not independent, but have variance dependencies is discussed. For this scenario Hyvärinen and Hurri (2004) proposed an algorithm which requires no assumption on distributions of sources and no parametric model of dependencies between components. In this paper, we extend the semiparametric approach of Amari and Cardoso (1997) to variance dependencies and study estimating functions for blind separation of such dependent sources. In particular, we show that many ICA algorithms are applicable to the variance-dependent model as well under mild conditions, although they should in principle not. Our results indicate that separation can be done based only on normalized sources which are adjusted to have stationary variances and is not affected by the dependent activity levels. We also study the asymptotic distribution of the quasi maximum likelihood method and the stability of the natural gradient learning in detail. Simulation results of artificial and realistic examples match well with our theoretical findings.

KW - Blind source separation

KW - Estimating functions

KW - Independent component analysis

KW - Semiparametric statistical models

KW - Variance dependencies

UR - http://www.scopus.com/inward/record.url?scp=21844458452&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=21844458452&partnerID=8YFLogxK

M3 - Article

VL - 6

JO - Journal of Machine Learning Research

JF - Journal of Machine Learning Research

SN - 1532-4435

ER -