Improving generalization performance in character recognition

Harris Drucker, Yann Le Cun

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

One test of a new training algorithm is how well the algorithm generalizes from the training data to the test data. A new training algorithm termed double backpropagation improves generalization by minimizing the change in the output due to small changes in the input. This is accomplished by minimizing the normal energy term found in backpropagation and an additional energy term that is a function of the Jacobian.

Original languageEnglish (US)
Title of host publicationNeural Networks for Signal Processing
PublisherPubl by IEEE
Pages198-207
Number of pages10
ISBN (Print)0780301188
StatePublished - 1991
EventProceedings of the 1991 Workshop on Neural Networks for Signal Processing - NNSP-91 - Princeton, NJ, USA
Duration: Sep 30 1991Oct 2 1991

Publication series

NameNeural Networks for Signal Processing

Other

OtherProceedings of the 1991 Workshop on Neural Networks for Signal Processing - NNSP-91
CityPrinceton, NJ, USA
Period9/30/9110/2/91

ASJC Scopus subject areas

  • Engineering(all)

Fingerprint Dive into the research topics of 'Improving generalization performance in character recognition'. Together they form a unique fingerprint.

  • Cite this

    Drucker, H., & Cun, Y. L. (1991). Improving generalization performance in character recognition. In Neural Networks for Signal Processing (pp. 198-207). (Neural Networks for Signal Processing). Publ by IEEE.