EEG Signal Classification using RBF Network Trained with Improved PSO Algorithm for Epilepsy Identification

Abstract

There are millions and billions of neurons present in a human brain, which are responsible for controlling human behaviour with internal or external stimuli response. These neurons considered as information carrier that passes the information between body and the brain. EEG signal got a very negligible amplitude and they are generated due to collision of neurons with each other. By reading these signals one can cure many brain disorder diseases. Epileptic seizure identification disease is one of them and it can be identified by doing the classification of EEG signal using Wavelet transform pre-processing and doing the classification using Radial Basis Function Neural Network. RBFNN is trained to optimize Mean Square Error by Improved Particle Swarm Optimization. Improved PSO is used to increase the searching in and around Global optimum solution. The procedure is checked on a benchmark publicly available dataset. Our experimental study with the help of results reveal that the improvement has been significant and remarkable over RBF trained by gradient descent and PSO.