Arab Handwriting Character Recognition Using Deep Learning

Abstract

Recent work has shown that neural networks have great potential in the field of handwriting recognition. The advantage of using this type of architecture, besides being robust, is that the network learns the characteristic vectors automatically thanks to the convolution layers. We can say that it creates intelligent filters. In this article we study deep learning in the field of Arab handwritten character in order to have a better understanding of its functioning. In this paper we present the work we have done on convolutional neural networks. First, we explain the theoretical aspects of neural networks, then we present our experimental protocols and we comment on the results obtained.