Header logo is

Using kernel PCA for Initialisation of Variational Bayesian Nonlinear Blind Source Separation Method

2004

Conference Paper

ei


The variational Bayesian nonlinear blind source separation method introduced by Lappalainen and Honkela in 2000 is initialised with linear principal component analysis (PCA). Because of the multilayer perceptron (MLP) network used to model the nonlinearity, the method is susceptible to local minima and therefore sensitive to the initialisation used. As the method is used for nonlinear separation, the linear initialisation may in some cases lead it astray. In this paper we study the use of kernel PCA (KPCA) in the initialisation. KPCA is a rather straightforward generalisation of linear PCA and it is much faster to compute than the variational Bayesian method. The experiments show that it can produce significantly better initialisations than linear PCA. Additionally, the model comparison methods provided by the variational Bayesian framework can be easily applied to compare different kernels.

Author(s): Honkela, A. and Harmeling, S. and Lundqvist, L. and Valpola, H.
Book Title: ICA 2004
Journal: Independent Component Analysis and Blind Signal Separation: Fifth International Conference (ICA 2004)
Pages: 790-797
Year: 2004
Month: October
Day: 0
Editors: Puntonet, C. G., A. Prieto
Publisher: Springer

Department(s): Empirical Inference
Bibtex Type: Conference Paper (inproceedings)

DOI: 10.1007/b100528
Event Name: Fifth International Conference on Independent Component Analysis and Blind Signal Separation
Event Place: Granada, Spain

Address: Berlin, Germany
Digital: 0
Language: en
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

BibTex

@inproceedings{6353,
  title = {Using kernel PCA for Initialisation of Variational Bayesian Nonlinear Blind Source Separation Method},
  author = {Honkela, A. and Harmeling, S. and Lundqvist, L. and Valpola, H.},
  journal = {Independent Component Analysis and Blind Signal Separation: Fifth International Conference (ICA 2004)},
  booktitle = {ICA 2004},
  pages = {790-797},
  editors = {Puntonet, C. G., A. Prieto},
  publisher = {Springer},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  address = {Berlin, Germany},
  month = oct,
  year = {2004},
  doi = {10.1007/b100528},
  month_numeric = {10}
}