Chinese Character Recognition

Neural Network

Neural Networks are powerful methods for classification. Our objective is to apply Neural Network (NN) in a character recognition problem to distinguish which character a given input is to represent its English equivalent. By teaching NN to adjust weights, we can find the relation between inputs and outputs. After that, we apply noisy characters which are never seen before to NN for classification. There are three main steps to solve this classification problem: Training, updating weights by using the differences between real output and desired output and the criterion to stop NN training.