Comparative analysis of machine learning algorithms in OCR

The purpose of this research is to implement different machine learning algorithms in optical character recognition. The algorithms used the pixel density of image of handwritten digits as an input. The algorithms when implemented produced the value of labels of each handwritten digit. The value of labels generated, was then matched with the actual value of labels of the MNIST handwritten digits to determine the accuracy of an algorithm. Machine learning algorithms that have been used for this research are Naïve Bayes, Naïve Bayes with Laplace Smoothing, Sequential Minimal Optimization, C4.5 decision trees and Logistic Regression. The accuracy for each of the algorithm was calculated and Logistic regression was found out to be the most accurate of them all for handwritten digits.