PSO learning on artificial neural networks

Abstract

The information processing capability of artificial neural networks (ANNs) is closely related to its architecture and connection weights. Various algorithms have been used to get an optimized structure suited for various applications. The paper does a comparative survey on effectiveness of evolving artificial neural networks based on the particle swarm optimization (PSO) algorithm over other algorithms.A description of this type of system tested on real problems in various domains has also been quoted. The results show that ANNs evolved by PSOANN have good accuracy and generalization ability.