Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Estimating functions for blind separation when sources have variance-dependencies

: Kawanabe, M.; Müller, K.-R.

Puntonet, C.G.:
Independent component analysis and blind signal separation : Fifth international conference, ICA 2004, Granada, Spain, September 22 - 24, 2004 ; proceedings
Berlin: Springer, 2004 (Lecture Notes in Computer Science 3195)
ISBN: 3-540-23056-4
ISSN: 0302-9743
International Conference on Independent Component Analysis and Blind Signal Seperation (ICA) <5, 2004, Granada>
Conference Paper
Fraunhofer FIRST ()

The blind separation problem where the sources are not independent, but have variance-dependencies is discussed. Hyvarinen and Hurri[1] 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 statistical approach of Amari and Cardoso[2] under variance-dependencies and study estimating functions for blind separation of such dependent sources. In particular, we show that many of ICA algorithms are applicable to the variance-dependent model as well. Our theoretical consequences were confirmed by artificial and realistic examples.