A variational approach to learning curves

Abstract

We combine the replica approach from statistical physics with a variational approach to analyze learning curves analytically. We apply the method to Gaussian process regression. As a main result we derive approximative relations between empirical error measures, the generalization error and the posterior variance.

Divisions: Aston University (General)
Additional Information: Copyright of Massachusetts Institute of Technology Press (MIT Press). Available from Google Scholar.
Uncontrolled Keywords: Gaussian process regression,generalization error,posterior variance
Publication ISSN: 1049-5258
Last Modified: 09 Jan 2024 08:04
Date Deposited: 09 Sep 2009 10:48
Full Text Link:
Related URLs: http://mitpress ... type=2&tid=8782 (Publisher URL)
PURE Output Type: Article
Published Date: 2002-09
Authors: Malzahn, Dorthe
Opper, Manfred

Download

[img]

Version: Accepted Version


Export / Share Citation


Statistics

Additional statistics for this record