Overcomplete ICA with a Geometric Algorithm

Fabian J. Theis, Elmar Wolfgang Lang, Tobias Westenhuber, Carlos García PuntonetICANN 2002 · OpenReview

Abstract

We present an independent component analysis (ICA) algorithm based on geometric considerations [10] [11] to decompose a linear mixture of more sources than sensor signals. Bofill and Zibulevsky [2] recently proposed a two-step approach for the separation: first learn the mixing matrix, then recover the sources using a maximum-likelihood approach. We present an efficient method for the matrix-recovery step mimicking the standard geometric algorithm thus generalizing Bofill and Zibulevsky’s method.
BibTeX
@inproceedings{theis2002overcomplete,
  author = {Fabian J. Theis and Elmar Wolfgang Lang and Tobias Westenhuber and Carlos García Puntonet},
  title = {{Overcomplete ICA with a Geometric Algorithm}},
  booktitle = {ICANN},
  year = {2002},
  url = {https://openreview.net/forum?id=iP6Bq7Tk4n_},
}

Loading related content...

Figure