Developing a system to predict handwriting numbers using ANN and CNN techniques

Abstract

The problem of recognizing handwritten numbers has become one of the most important problems in machine learning and computer vision applications. Many machine learning techniques have been used to solve the problem of recognizing handwritten numbers. In this paper, a system was developed that distinguishes handwritten numbers using two machine learning algorithms, which are Artificial Neural networks (ANN) and Convolutional Neural networks (CNN). A special database for handwritten numbers, the Modified National Institute of Standards and Technology, was used. Database (MNIST). The results were compared based on accuracy and performance with the Batch Normalization and Dropout layers. The results were good. The CNN obtained a score of 99% with the use of BN+PO compared to the ANN in distinguishing handwritten numbers.