Improving generalization performance in character recognition

Harris Drucker, Yann LeCun

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

Other

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

Fingerprint

Character recognition
Backpropagation

ASJC Scopus subject areas

  • Engineering(all)

Cite this

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

Improving generalization performance in character recognition. / Drucker, Harris; LeCun, Yann.

Neural Networks for Signal Processing. Publ by IEEE, 1991. p. 198-207.

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

Drucker, H & LeCun, Y 1991, Improving generalization performance in character recognition. in Neural Networks for Signal Processing. Publ by IEEE, pp. 198-207, Proceedings of the 1991 Workshop on Neural Networks for Signal Processing - NNSP-91, Princeton, NJ, USA, 9/30/91.
Drucker H, LeCun Y. Improving generalization performance in character recognition. In Neural Networks for Signal Processing. Publ by IEEE. 1991. p. 198-207
Drucker, Harris ; LeCun, Yann. / Improving generalization performance in character recognition. Neural Networks for Signal Processing. Publ by IEEE, 1991. pp. 198-207
@inproceedings{87e7a99f67ad4f5387813dbcbe17b3be,
title = "Improving generalization performance in character recognition",
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.",
author = "Harris Drucker and Yann LeCun",
year = "1991",
language = "English (US)",
isbn = "0780301188",
pages = "198--207",
booktitle = "Neural Networks for Signal Processing",
publisher = "Publ by IEEE",

}

TY - GEN

T1 - Improving generalization performance in character recognition

AU - Drucker, Harris

AU - LeCun, Yann

PY - 1991

Y1 - 1991

N2 - 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.

AB - 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.

UR - http://www.scopus.com/inward/record.url?scp=0026289340&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0026289340&partnerID=8YFLogxK

M3 - Conference contribution

SN - 0780301188

SP - 198

EP - 207

BT - Neural Networks for Signal Processing

PB - Publ by IEEE

ER -