General Regression Neural Network Optimization for Handwritten Persian Digits recognition Using Particle Swarm Optimization

Abstract

In this paper an optimization algorithm based on Particle Swarm Optimization algorithm is used for handwritten Persian digits recognition with General Regression Neural Network .The system uses image zoning for the digit recognition. General Regression Neural Network accuracy depends on the centers and widths of the hidden layer neuron basis functions (neuron spread). In this paper we use Particle Swarm Optimization algorithm to determine General Regression Neural Network hidden layer spread. Results show that the optimized General Regression Neural Network provides higher recognition ability compared with that of unoptimized General Regression Neural Network.