Word-level training of a handritten word recognizer based on convolutional neural networks

Yann Le Cun; Yoshua Bengio

Word-level training of a handritten word recognizer based on convolutional neural networks

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. Words are represented by low resolution "annotated images" where each pixel contains information about trajectory direction and curvature. The recognizer is a convolutional network which can be spatially replicated. From the network output, a hidden Markov model produces word scores. The entire system is globally trained to minimize word-level errors.

Original language	English (US)
Title of host publication	Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B
Subtitle of host publication	Pattern Recognition and Neural Networks, ICPR 1994
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	88-92
Number of pages	5
ISBN (Electronic)	0818662700
State	Published - 1994
Event	12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994 - Jerusalem, Israel Duration: Oct 9 1994 → Oct 13 1994

Publication series

Name	Proceedings - International Conference on Pattern Recognition
Volume	2
ISSN (Print)	1051-4651

Conference

Conference	12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994
Country/Territory	Israel
City	Jerusalem
Period	10/9/94 → 10/13/94

ASJC Scopus subject areas

Computer Vision and Pattern Recognition

Cite this

Cun, Y. L., & Bengio, Y. (1994). Word-level training of a handritten word recognizer based on convolutional neural networks. In Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994 (pp. 88-92). (Proceedings - International Conference on Pattern Recognition; Vol. 2). Institute of Electrical and Electronics Engineers Inc..

Word-level training of a handritten word recognizer based on convolutional neural networks. / Cun, Yann Le; Bengio, Yoshua.
Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994. Institute of Electrical and Electronics Engineers Inc., 1994. p. 88-92 (Proceedings - International Conference on Pattern Recognition; Vol. 2).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Cun, YL & Bengio, Y 1994, Word-level training of a handritten word recognizer based on convolutional neural networks. in Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994. Proceedings - International Conference on Pattern Recognition, vol. 2, Institute of Electrical and Electronics Engineers Inc., pp. 88-92, 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994, Jerusalem, Israel, 10/9/94.

Cun YL, Bengio Y. Word-level training of a handritten word recognizer based on convolutional neural networks. In Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994. Institute of Electrical and Electronics Engineers Inc. 1994. p. 88-92. (Proceedings - International Conference on Pattern Recognition).

Cun, Yann Le ; Bengio, Yoshua. / Word-level training of a handritten word recognizer based on convolutional neural networks. Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994. Institute of Electrical and Electronics Engineers Inc., 1994. pp. 88-92 (Proceedings - International Conference on Pattern Recognition).

@inproceedings{1de047738c5f44098dec6243c1d9d44e,

title = "Word-level training of a handritten word recognizer based on convolutional neural networks",

abstract = "We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. Words are represented by low resolution {"}annotated images{"} where each pixel contains information about trajectory direction and curvature. The recognizer is a convolutional network which can be spatially replicated. From the network output, a hidden Markov model produces word scores. The entire system is globally trained to minimize word-level errors.",

author = "Cun, {Yann Le} and Yoshua Bengio",

note = "Publisher Copyright: {\textcopyright} 1994 Institute of Electrical and Electronics Engineers Inc.. All rights reserved.; 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994 ; Conference date: 09-10-1994 Through 13-10-1994",

year = "1994",

language = "English (US)",

series = "Proceedings - International Conference on Pattern Recognition",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "88--92",

booktitle = "Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B",

}

TY - GEN

T1 - Word-level training of a handritten word recognizer based on convolutional neural networks

AU - Cun, Yann Le

AU - Bengio, Yoshua

PY - 1994

Y1 - 1994

N2 - We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. Words are represented by low resolution "annotated images" where each pixel contains information about trajectory direction and curvature. The recognizer is a convolutional network which can be spatially replicated. From the network output, a hidden Markov model produces word scores. The entire system is globally trained to minimize word-level errors.

AB - We introduce a new approach for on-line recognition of handwritten words written in unconstrained mixed style. Words are represented by low resolution "annotated images" where each pixel contains information about trajectory direction and curvature. The recognizer is a convolutional network which can be spatially replicated. From the network output, a hidden Markov model produces word scores. The entire system is globally trained to minimize word-level errors.

UR - http://www.scopus.com/inward/record.url?scp=85093199842&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85093199842&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85093199842

T3 - Proceedings - International Conference on Pattern Recognition

SP - 88

EP - 92

BT - Proceedings of the 12th IAPR International Conference on Pattern Recognition - Conference B

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 12th IAPR International Conference on Pattern Recognition - Conference B: Pattern Recognition and Neural Networks, ICPR 1994

Y2 - 9 October 1994 through 13 October 1994

ER -

Word-level training of a handritten word recognizer based on convolutional neural networks

Abstract

Publication series

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this