Title :
Multilayer perceptron and uppercase handwritten characters recognition
Author :
Bernard, Ir Gosselin
Author_Institution :
Service de Theorie des Circuits et de Traitement du Signal, Fac. Polytech. de Mons, Belgium
Abstract :
After an introduction to the problem of the automatic character recognition and on multilayer perceptron used for classification, the author describes what one can hope to get from a multilayer perceptron. Some of the problems that can occur during the training and present a fast learning algorithm are also described. This algorithm was tested to train a multilayer perceptron to recognize multiscriptor uppercase handwritten characters. The system has reached a recognition rate of 88.1%, without any contextual analysis, which is still indispensable, but will be easier due to the fact that the multilayer perceptron provides the probability of each class to be the unknown character
Keywords :
character recognition; handwriting recognition; learning systems; multilayer perceptrons; pattern classification; automatic character recognition; classification; contextual analysis; fast learning algorithm; multilayer perceptron; multiscriptor uppercase handwritten characters; training; uppercase handwritten characters recognition; Character recognition; Circuits; Handwriting recognition; High performance computing; Multi-layer neural network; Multilayer perceptrons; Neural networks; Neurons; Performance analysis; Testing;
Conference_Titel :
Document Analysis and Recognition, 1993., Proceedings of the Second International Conference on
Conference_Location :
Tsukuba Science City
Print_ISBN :
0-8186-4960-7
DOI :
10.1109/ICDAR.1993.395583
