Recognition System of Hand-written Figures by Using Neural Networks and Genetic Algorithm

This paper describes the development of the computer system to recognize hand-written ten figures from 0 to 9 by learning synaptic weights of neural networks including genetic algorithm. The genetic algorithm process is (1) To generate initial population, (2) To calculate fitness for each individual based on the recognition results of hand-written teaching data, (3) Genetic operations such as crossover and mutation, (4) To select high fitness individuals and (5) To judge conditions. After the recognition simulations, it is ascertained that the developed system has the highest recognition rate.