Handwritten Digit Recognition Based on Principal Component Analysis and Support Vector Machines

Handwritten digit recognition has always been a challenging task in pattern recognition area. In this paper we explore the performance of support vector machines (SVM) and principal component analysis (PCA) on handwritten digits recognition. The performance of SVM on handwritten digits recognition task is compared with three typical classification methods, i.e., linear discriminant classifiers (LDC), the nearest neighbor (1-NN), and the back-propagation neural network (BPNN). The experimental results on the popular MNIST database indicate that SVM gets the best performance with an accuracy of 89.7% with 10-dimensional embedded features, outperforming the other used methods.