Developing a New CCN Technique for Arabic Handwritten Digits Recognition

Convolutional Neural Networks (CNN) have many applications in object recognition such as character and digit recognition. Few researches are performed on Arabic handwritten digits recognition. In this research, we will develop a new algorithm to utilize the convolutional neural networks with sigmoid function (σ-CNN) to recognize Arabic handwritten digits recognition. The performance of this method provides minimum cost functions with maximum testing accuracy results in compared to other existing techniques