Associative memories for handwritten pattern recognition

Author

L?³pez-Aligu?©, Francisco J. ; Acevedo-Sotoca, Isabel ; Orellana, Carlos G. ; Velasco, Horacio G.

Author_Institution

Dept. of Electron. & Electrochem. Eng., Univ. de Extremadura, Badajoz, Spain

Volume

2

fYear

2003

Firstpage

1441

Abstract

We describe the construction of a complete system of handwritten classification on the basis of bidirectional associative memories (BAM). The BAM is synthesized with a two-layer neural network in which the neuron thresholds are adjusted at the moment of learning, and the ideal prototypes for each class are selected using an automatic method. The system is completed with a topological preprocessor that eliminates distortion and noise components. It was tested on the popular NIST#19 database, attaining 100% success rates for all the characters, even under conditions of high levels of contamination due to noise or distortion of the input image.

Keywords

content-addressable storage; handwritten character recognition; image denoising; neural nets; NIST#19 database; bidirectional associative memories; distortion; handwritten pattern recognition; noise; topological preprocessor; two-layer neural network; Associative memory; Contamination; Image databases; Magnesium compounds; Network synthesis; Neural networks; Neurons; Pattern recognition; Prototypes; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 2003. Proceedings of the International Joint Conference on

ISSN

1098-7576

Print_ISBN

0-7803-7898-9

Type

conf

DOI

10.1109/IJCNN.2003.1223908

Filename

1223908