Blind Source Separation for Non-Stationary Mixing

  • Richard Everson
  • Stephen Roberts
Article

Abstract

Blind source separation attempts to recover independent sources which have been linearly mixed to produce observations. We consider blind source separation with non-stationary mixing, but stationary sources. The linear mixing of the independent sources is modelled as evolving according to a Markov process, and a method for tracking the mixing and simultaneously inferring the sources is presented. Observational noise is included in the model. The technique may be used for online filtering or retrospective smoothing. The tracking of mixtures of temporally correlated is examined and sampling from within a sliding window is shown to be effective for destroying temporal correlations. The method is illustrated with numerical examples.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Kluwer Academic Publishers 2000

Authors and Affiliations

  • Richard Everson
    • 1
  • Stephen Roberts
    • 2
  1. 1.Department of Computer ScienceUniversity of ExeterExeterUK
  2. 2.Department of Engineering ScienceUniversity of OxfordOxfordUK

Personalised recommendations