Statistical Theory of Overtraining - Is Cross-Validation Asymptotically Effective?

  • S. Amari
  • , N. Murata
  • , K. R. Müller
  • , M. Finke
  • , H. Yang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

45 Scopus citations

Abstract

A statistical theory for overtraining is proposed. The analysis treats realizable stochastic neural networks, trained with Kullback-Leibler loss in the asymptotic case. It is shown that the asymptotic gain in the generalization error is small if we perform early stopping, even if we have access to the optimal stopping time. Considering cross-validation stopping we answer the question: In what ratio the examples should be divided into training and testing sets in order to obtain the optimum performance. In the non-asymptotic region cross-validated early stopping always decreases the generalization error. Our large scale simulations done on a CM5 are in nice agreement with our analytical findings.

Original languageEnglish
Title of host publicationAdvances in Neural Information Processing Systems 8, NIPS 1995
EditorsD. Touretzky, M.C. Mozer, M. Hasselmo
PublisherNeural information processing systems foundation
Pages176-182
Number of pages7
ISBN (Electronic)0262201070, 9780262201070
StatePublished - 1995
Externally publishedYes
Event8th Advances in Neural Information Processing Systems, NIPS 1995 - Denver, United States
Duration: 27 Nov 199530 Nov 1995

Publication series

NameAdvances in Neural Information Processing Systems
Volume8
ISSN (Print)1049-5258

Conference

Conference8th Advances in Neural Information Processing Systems, NIPS 1995
Country/TerritoryUnited States
CityDenver
Period27/11/9530/11/95

Fingerprint

Dive into the research topics of 'Statistical Theory of Overtraining - Is Cross-Validation Asymptotically Effective?'. Together they form a unique fingerprint.

Cite this