Handwritten digit recognition: applications of neural network chips and automatic learning

IEEE Communications MagazinePublished 1 November 1989

Y. Le Cun, L. D. Jackel, Bernhard E. Boser, J. S. Denker, Hans Peter Graf, Isabelle Guyon

DOI10.1109/35.41400

Source URL

Citations470

SJR quartileQ1

SJR score3.28

SNIP2.80

TL;DR

Two novel methods for achieving handwritten digit recognition are described, based on a neural network chip that performs line thinning and feature extraction using local template matching and on a digital signal processor that makes extensive use of constrained automatic learning.

Abstract

Two novel methods for achieving handwritten digit recognition are described. The first method is based on a neural network chip that performs line thinning and feature extraction using local template matching. The second method is implemented on a digital signal processor and makes extensive use of constrained automatic learning. Experimental results obtained using isolated handwritten digits taken from postal zip codes, a rather difficult data set, are reported and discussed.>

Keywords

Computer ScienceEngineering

Learning Internal Representations by Error Propagation

16,235 Citations1985David E. Rumelhart, Geoffrey E. Hinton +1 more

Connectionism in Perspective

509 Citations1989Rolf Pfeifer, Françoise Fogelman-Soulie +2 more

IEEE Transactions on Systems Man and CyberneticsSPTA: A proposed algorithm for thinning binary patterns

219 Citations1984Nabil Jean Naccache, Rajjan Shinghal

A new skeletonization algorithm called the safe-point thinning algorithm (SPTA) is proposed that is suitable for binary patterns and is shown to be the fastest and that it produces skeletons of good quality.

Consistent inference of probabilities in layered networks: predictions and generalizations

157 Citations1989Tishby, Левин +1 more

The problem of learning a general input-output relation using a layered neural network is discussed in a statistical framework and the authors arrive at a Gibbs distribution on a canonical ensemble of networks with the same architecture.

Neural Network Recognizer for Hand-Written Zip Code Digits

150 Citations1988John S. Denker, W.R. Gardner +7 more

This paper describes the construction of a system that recognizes handprinted digits, using a combination of classical techniques and neural-net methods, and compares favorably with other state-of-the art recognizers.

IBM Journal of Research and DevelopmentMoment Normalization of Handprinted Characters

123 Citations1970R. G. Casey

Comparison experiments showed that error rates were reduced by integral factors if the patterns were normalized before scanning for recognition, and second-order moments of the pattern are convenient properties to use in specifying the transformation.

Europhysics Letters (EPL)Learning Networks of Neurons with Boolean Logic

70 Citations1987Stefano Patarnello, P. Carnevali

A model which self-organizes to perform a task via a learning-by-example scheme which is a network of Boolean operators which has been able to achieve an error-free design for addition between integer binary numbers when shown only a small subset of all possible additions.

A CMOS associative memory chip based on neural networks

48 Citations1987Hans Peter Graf, P. de Vegvar

A chip containing 54 amplifiers, 6K SRAM and programmable interconnections, that has been used to implement an algorithm based on biological neural networks, and ten vectors stored in the memory may be recalled within 500ns.

IEEE Transactions on ComputersSome Studies in the Interactive Design of Character Recognition Systems

13 Citations1971W. Clark Naylor

The development of more powerful multilevel logic forms and suggestions for making these forms easy to use in an interactive design environment are particularly important results of these studies.