Abstract: In order to recognise the hand written character the system model is developed which is based upon neural network. MATLAB software including neural network toll box is used. The attributes and advantages which ultimately achieve the importance of design model with greatest ability. To interpret the meaning of complex or imprecise data available which is use to recognise the pattern and detect very easily which inherently difficult for human being and to any computer technique .In this technique the neural network is treated as expert which is use to analyse. Study is carried out using artificial neural network (ANN) which is an information processing model that is motivated by biological nervous system works such as human brain processes information. The comparative study to recognise the character is done using multilayer perceptron neural network (MLP), support vector machine (SPM), radial basis function (RBF) After studying it is found that percentage accuracy obtain in support vector machine is 90-100% and main square error (MSE) is least the, percent accuracy obtain in multilayer perceptron neural network is near about 80-90% and MSE is less so SVMNN is the best classifier so far as hand written character recognition system is concerned. MLPNN is slightly inferior in performance as compared to the SVMNN.

Keywords: Neural network: SVM, MLP, RBF, MATLAB, JPG images of 20 person’s hand written character