A Study on the Application of Pattern Recognition in Handwritten Numerical Recognition

This paper analyzes the problems of handwritten numerical recognition, and uses Sequential Minimal Optimization (SMO) to train and recognize the character data of handwritten numeral which are gathered from many different people. Handwritten recognition, which has become one of the important methods in information gathering and information transformation, has a promising feature in many fields in society. Compared with neural network, this paper finds that Support Vector Machine (SVM) has the better recognition rate.