Missing Data in Kernel PCA

Guido Sanguinetti; Neil D. Lawrence

edit

Back to publications

Missing Data in Kernel PCA

Guido Sanguinetti, Neil D. Lawrence

ECML, Berlin, 2006, Springer-Verlag :751-758, 2006.

Abstract

Kernel Principal Component Analysis (KPCA) is a widely used technique for visualisation and feature extraction. Despite its success and flexibility, the lack of a probabilistic interpretation means that some problems, such as handling missing or corrupted data, are very hard to deal with. In this paper we exploit the probabilistic interpretation of linear PCA together with recent results on latent variable models in Gaussian Processes in order to introduce an objective function for KPCA. This in turn allows a principled approach to the missing data problem. Furthermore, this new approach can be extended to reconstruct corrupted test data using fixed kernel feature extractors. The experimental results show strong improvements over widely used heuristics.

Links

Cite this Paper

BibTeX


@InProceedings{Sanguinetti:missingkpca06,
  title = 	 {Missing Data in Kernel {PCA}},
  author = 	 {Sanguinetti, Guido and Lawrence, Neil D.},
  booktitle = 	 {ECML, Berlin, 2006},
  pages = 	 {751--758},
  year = 	 {2006},
  series = 	 {Lecture Notes in Computer Science},
  address = 	 {Berlin},
  publisher =    {Springer-Verlag},
  pdf = 	 {https://inverseprobability.com/publications/files/CS0608.pdf},
  url = 	 {/publications/sanguinetti-missingkpca06.html},
  abstract = 	 {Kernel Principal Component Analysis (KPCA) is a widely used
technique for visualisation and feature extraction. Despite its
success and flexibility, the lack of a probabilistic interpretation
means that some problems, such as handling missing or corrupted
data, are very hard to deal with. In this paper we exploit the
probabilistic interpretation of linear PCA together with recent
results on latent variable models in Gaussian Processes in order to
introduce an objective function for KPCA. This in turn allows a
principled approach to the missing data problem. Furthermore, this
new approach can be extended to reconstruct corrupted test data
using fixed kernel feature extractors. The experimental results show
strong improvements over widely used heuristics.
}
}

Endnote

%0 Conference Paper
%T Missing Data in Kernel PCA
%A Guido Sanguinetti
%A Neil D. Lawrence
%B ECML, Berlin, 2006
%C Lecture Notes in Computer Science
%D 2006	
%F Sanguinetti:missingkpca06
%I Springer-Verlag
%P 751--758
%U /publications/sanguinetti-missingkpca06.html
%X Kernel Principal Component Analysis (KPCA) is a widely used
technique for visualisation and feature extraction. Despite its
success and flexibility, the lack of a probabilistic interpretation
means that some problems, such as handling missing or corrupted
data, are very hard to deal with. In this paper we exploit the
probabilistic interpretation of linear PCA together with recent
results on latent variable models in Gaussian Processes in order to
introduce an objective function for KPCA. This in turn allows a
principled approach to the missing data problem. Furthermore, this
new approach can be extended to reconstruct corrupted test data
using fixed kernel feature extractors. The experimental results show
strong improvements over widely used heuristics.

RIS


TY  - CPAPER
TI  - Missing Data in Kernel PCA
AU  - Guido Sanguinetti
AU  - Neil D. Lawrence
BT  - ECML, Berlin, 2006
DA  - 2006/06/19	
ID  - Sanguinetti:missingkpca06
PB  - Springer-Verlag
DP  - Lecture Notes in Computer Science
SP  - 751
EP  - 758
L1  - https://inverseprobability.com/publications/files/CS0608.pdf
UR  - /publications/sanguinetti-missingkpca06.html
AB  - Kernel Principal Component Analysis (KPCA) is a widely used
technique for visualisation and feature extraction. Despite its
success and flexibility, the lack of a probabilistic interpretation
means that some problems, such as handling missing or corrupted
data, are very hard to deal with. In this paper we exploit the
probabilistic interpretation of linear PCA together with recent
results on latent variable models in Gaussian Processes in order to
introduce an objective function for KPCA. This in turn allows a
principled approach to the missing data problem. Furthermore, this
new approach can be extended to reconstruct corrupted test data
using fixed kernel feature extractors. The experimental results show
strong improvements over widely used heuristics.

ER  -

APA


Sanguinetti, G. & Lawrence, N.D.. (2006). Missing Data in Kernel PCA. ECML, Berlin, 2006, in Lecture Notes in Computer Science:751-758 Available from /publications/sanguinetti-missingkpca06.html.