Recognition of unconstrained handwritten digits using modified chaotic neural networks

Abstract

This paper describes an off-line method for recognizing totally unconstrained handwritten digits using modified chaotic neural networks (CNN). Since the CNN has inherently the characteristics of highly nonlinear dynamics it can be an appropriate network for the robust classification of complex patterns. The CNN in this paper is trained by the error backpropagation algorithm. Digit identification starts with extraction of features from the raw digit images and then recognizes digits using the CNN based classifier The performance of the CNN classifier is evaluated on the Concordia database. For the relative comparison of recognition performance the CNN classifier is compared with the recurrent neural networks (RNN) classifier Experimental results show that the classification rate is 98.4%. It indicates that the CNN classifier outperforms the RNN classifier as well as other classifiers that have been reported on the same database.