Sparse deflations in blind signal separation

Abstract

We present a new deflation procedure for blind signal separation based on sparsity. It allows, under mild sparsity assumptions, to separate mixtures which could not be separated by ICA methods. We present a new algorithm for sparse deflations and apply it for sparse blind signal separation of mixtures of signals with bounded support. Relations to signals from High Performance Liquid Chromatography in chemistry are discussed and computer simulation examples are presented. © Springer-Verlag Berlin Heidelberg 2006.