Empirical Inference

Brisk Kernel ICA

2007

Book Chapter

ei


Recent approaches to independent component analysis have used kernel independence measures to obtain very good performance in ICA, particularly in areas where classical methods experience difficulty (for instance, sources with near-zero kurtosis). In this chapter, we compare two efficient extensions of these methods for large-scale problems: random subsampling of entries in the Gram matrices used in defining the independence measures, and incomplete Cholesky decomposition of these matrices. We derive closed-form, efficiently computable approximations for the gradients of these measures, and compare their performance on ICA using both artificial and music data. We show that kernel ICA can scale up to much larger problems than yet attempted, and that incomplete Cholesky decomposition performs better than random sampling.

Author(s): Jegelka, S. and Gretton, A.
Book Title: Large Scale Kernel Machines
Pages: 225-250
Year: 2007
Month: September
Day: 0

Series: Neural Information Processing
Editors: Bottou, L. , O. Chapelle, D. DeCoste, J. Weston
Publisher: MIT Press

Department(s): Empirical Inference
Bibtex Type: Book Chapter (inbook)

Address: Cambridge, MA, USA
Digital: 0
Language: en
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

Links: PDF
Web

BibTex

@inbook{4192,
  title = {Brisk Kernel ICA},
  author = {Jegelka, S. and Gretton, A.},
  booktitle = {Large Scale Kernel Machines},
  pages = {225-250},
  series = {Neural Information Processing},
  editors = {Bottou, L. , O. Chapelle, D. DeCoste, J. Weston},
  publisher = {MIT Press},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  address = {Cambridge, MA, USA},
  month = sep,
  year = {2007},
  doi = {},
  month_numeric = {9}
}