Abstract

In this modern world, technology plays an vital role to reduce the man power. Costly\nmanual labour is required to do a tedious job of converting the physical written data and\ninformation into digital form. Handwritten digit recognition is a methodology that\nautomatically recognizing and detecting handwritten digital data through different Machine\nLearning models .In this project we use hybrid approach of machine learning algorithms to\nenhance the productiveness of technique and reduce the complexity of using various\nmodels.Handwritten Digit Recognition is a pivotal concern in computer vision.We mainly\nfocused on performance and consistency of machine learning algorithms through hybrid\napproach.It will be more efficiency and accuracy for any digital data sets compared to the\nexisting system.We have used the MNIST dataset for training and recognition which consist of\nset for hand written digits (0-9).The data consist of 70,000 images for training and testing. Each\ndigit is represented as a 28 by 28 grey scale pixels i.e. 784 pixels intensities for better results.