HAPSOENN: Hybrid Accelerated Particle Swarm Optimized Elman Neural Network

Abstract

Back propagation (BP) algorithm is a very popular optimization procedure of ANN’s training process. However, traditional BP has some drawbacks such as getting stuck in local minima, and network stagnancy. Recently, some researches proposed the use of Elman Neural Network (ENN) trained with back propagation algorithm to yield faster and more accurate results during learning. Yet, the performance of ENN is still considerably dependent on initial weights in the network. In this paper, a new method known as HAPSOENN which adapts the network weights using Accelerated Particle Swarm Optimization (APSO) was proposed as a mechanism to improve the performance of ENN. The performance of the proposed algorithm is compared with Back-Propagation Neural Network (BPNN) and other similar hybrid variants on benchmarked classification datasets. The simulation results show that the proposed technique performs better and has faster convergence than other algorithms in terms of MSE and accuracy.