Header logo is

A new discriminative kernel from probabilistic models

2002

Conference Paper

ei


Recently, Jaakkola and Haussler proposed a method for constructing kernel functions from probabilistic models. Their so called \Fisher kernel" has been combined with discriminative classi ers such as SVM and applied successfully in e.g. DNA and protein analysis. Whereas the Fisher kernel (FK) is calculated from the marginal log-likelihood, we propose the TOP kernel derived from Tangent vectors Of Posterior log-odds. Furthermore, we develop a theoretical framework on feature extractors from probabilistic models and use it for analyzing the TOP kernel. In experiments our new discriminative TOP kernel compares favorably to the Fisher kernel.

Author(s): Tsuda, K. and Kawanabe, M. and Rätsch, G. and Sonnenburg, S. and Müller, K-R.
Book Title: Advances in Neural Information Processing Systems 14
Journal: Advances in Neural Information Processing Systems
Pages: 977-984
Year: 2002
Month: September
Day: 0
Editors: Dietterich, T.G. , S. Becker, Z. Ghahramani
Publisher: MIT Press

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

Event Name: Fifteenth Annual Neural Information Processing Systems Conference (NIPS 2001)
Event Place: Vancouver, BC, Canada

Address: Cambridge, MA, USA
Digital: 0
ISBN: 0-262-04208-8
Organization: Max-Planck-Gesellschaft
School: Biologische Kybernetik

Links: PDF
Web

BibTex

@inproceedings{2191,
  title = {A new discriminative kernel from probabilistic models},
  author = {Tsuda, K. and Kawanabe, M. and R{\"a}tsch, G. and Sonnenburg, S. and M{\"u}ller, K-R.},
  journal = {Advances in Neural Information Processing Systems},
  booktitle = {Advances in Neural Information Processing Systems 14},
  pages = {977-984},
  editors = {Dietterich, T.G. , S. Becker, Z. Ghahramani},
  publisher = {MIT Press},
  organization = {Max-Planck-Gesellschaft},
  school = {Biologische Kybernetik},
  address = {Cambridge, MA, USA},
  month = sep,
  year = {2002},
  month_numeric = {9}
}