Object recognition with gradient-based learning

Yann LeCun, Patrick Haffner, Léon Bottou, Yoshua Bengio

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

Abstract

Finding an appropriate set of features s an essential problem in the design of shape recognition systems. This paper attempts to show that for recognizing simple objects with high shape variability such as handwritten characters, it is possible, and even advantageous, to feed the system directly with minimally processed images and to rely on learning to extract the right set of features. Convolutional Neural Networks are shown to be particularly well suited to this task. We also show that these networks can be used to recognize multiple objects without requiring explicit segmentation of the objects from their surrounding. The second part of the paper presents the Graph Transformer Network model which extends the applicability of gradient-based learning to systems that use graphs to represents features, objects, and their combinations.

Original languageEnglish (US)
Title of host publicationShape, Contour and Grouping in Computer Vision
PublisherSpringer Verlag
Pages319-345
Number of pages27
Volume1681
ISBN (Print)3540667229, 9783540667223
StatePublished - 1999
EventInternational Workshop on Shape, Contour and Grouping in Computer Vision - Palermo, Sicily, Italy
Duration: May 26 1998May 29 1998

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume1681
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

OtherInternational Workshop on Shape, Contour and Grouping in Computer Vision
CountryItaly
CityPalermo, Sicily
Period5/26/985/29/98

Fingerprint

Object recognition
Object Recognition
Gradient
Neural networks
Shape Recognition
Transformer
Graph in graph theory
Network Model
Segmentation
Neural Networks
Learning
Object

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

LeCun, Y., Haffner, P., Bottou, L., & Bengio, Y. (1999). Object recognition with gradient-based learning. In Shape, Contour and Grouping in Computer Vision (Vol. 1681, pp. 319-345). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1681). Springer Verlag.

Object recognition with gradient-based learning. / LeCun, Yann; Haffner, Patrick; Bottou, Léon; Bengio, Yoshua.

Shape, Contour and Grouping in Computer Vision. Vol. 1681 Springer Verlag, 1999. p. 319-345 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1681).

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

LeCun, Y, Haffner, P, Bottou, L & Bengio, Y 1999, Object recognition with gradient-based learning. in Shape, Contour and Grouping in Computer Vision. vol. 1681, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1681, Springer Verlag, pp. 319-345, International Workshop on Shape, Contour and Grouping in Computer Vision, Palermo, Sicily, Italy, 5/26/98.
LeCun Y, Haffner P, Bottou L, Bengio Y. Object recognition with gradient-based learning. In Shape, Contour and Grouping in Computer Vision. Vol. 1681. Springer Verlag. 1999. p. 319-345. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
LeCun, Yann ; Haffner, Patrick ; Bottou, Léon ; Bengio, Yoshua. / Object recognition with gradient-based learning. Shape, Contour and Grouping in Computer Vision. Vol. 1681 Springer Verlag, 1999. pp. 319-345 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{3a6e6cc3ac6e4326b35a6315d7cbddf7,
title = "Object recognition with gradient-based learning",
abstract = "Finding an appropriate set of features s an essential problem in the design of shape recognition systems. This paper attempts to show that for recognizing simple objects with high shape variability such as handwritten characters, it is possible, and even advantageous, to feed the system directly with minimally processed images and to rely on learning to extract the right set of features. Convolutional Neural Networks are shown to be particularly well suited to this task. We also show that these networks can be used to recognize multiple objects without requiring explicit segmentation of the objects from their surrounding. The second part of the paper presents the Graph Transformer Network model which extends the applicability of gradient-based learning to systems that use graphs to represents features, objects, and their combinations.",
author = "Yann LeCun and Patrick Haffner and L{\'e}on Bottou and Yoshua Bengio",
year = "1999",
language = "English (US)",
isbn = "3540667229",
volume = "1681",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "319--345",
booktitle = "Shape, Contour and Grouping in Computer Vision",

}

TY - GEN

T1 - Object recognition with gradient-based learning

AU - LeCun, Yann

AU - Haffner, Patrick

AU - Bottou, Léon

AU - Bengio, Yoshua

PY - 1999

Y1 - 1999

N2 - Finding an appropriate set of features s an essential problem in the design of shape recognition systems. This paper attempts to show that for recognizing simple objects with high shape variability such as handwritten characters, it is possible, and even advantageous, to feed the system directly with minimally processed images and to rely on learning to extract the right set of features. Convolutional Neural Networks are shown to be particularly well suited to this task. We also show that these networks can be used to recognize multiple objects without requiring explicit segmentation of the objects from their surrounding. The second part of the paper presents the Graph Transformer Network model which extends the applicability of gradient-based learning to systems that use graphs to represents features, objects, and their combinations.

AB - Finding an appropriate set of features s an essential problem in the design of shape recognition systems. This paper attempts to show that for recognizing simple objects with high shape variability such as handwritten characters, it is possible, and even advantageous, to feed the system directly with minimally processed images and to rely on learning to extract the right set of features. Convolutional Neural Networks are shown to be particularly well suited to this task. We also show that these networks can be used to recognize multiple objects without requiring explicit segmentation of the objects from their surrounding. The second part of the paper presents the Graph Transformer Network model which extends the applicability of gradient-based learning to systems that use graphs to represents features, objects, and their combinations.

UR - http://www.scopus.com/inward/record.url?scp=84947927646&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84947927646&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:84947927646

SN - 3540667229

SN - 9783540667223

VL - 1681

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 319

EP - 345

BT - Shape, Contour and Grouping in Computer Vision

PB - Springer Verlag

ER -