2012 | OriginalPaper | Chapter
Handwritten Character Recognition with Artificial Neural Networks
Authors : Stephane Kouamo, Claude Tangha
Published in: Distributed Computing and Artificial Intelligence
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Apart from differents techniques studied in an increasing order of difficulty, this work presents a new approach based on the use of a multilayer perceptron with the optimal hidden neurons. Idea is to compute the training stage by using two classes of prototypes, to represent data already known ; hidden layers are then initialized by that two classes of prototypes. One of the advantages of this technique is the use of the second hiden layer which allows the network to filter better the case of nearby data. The results come from the
Yann Le Cun
database [9], and show that the approach based on the use of a multilayer perceptron with two hidden layers is very promising, though improvable.