EEG Based Feature Extraction & ANN Classifier’s Neurons Selection for Epileptic Classification Using Hybrid Soft Computing (ANN & GA)

Authors

  • Manoj Kumar Bandil, Dr. A. K. Wadhwani

Abstract

Nervous system cells including neurons and neurons or nerve cells transmit information and non-neuronal cells from the brain.  They organize themselves into complex  network that performs nervous system functions.. Execute half breed processing strategy utilizing Genetic Algorithms (GA) and Artificial Neural Networks (ANN) for determination of features and no of neurons in the shrouded layers of ANN classifier to improve classification exactness for epileptic cases. Looking at epileptic classifications results for single concealed layer and twofold shrouded layer ANN classifier after determination of features and no of neurons in the concealed layer by utilizing half and half calculation presumed that twofold concealed layer ANN classifier shows improved precision in arranging EEG signals.The main purpose of our research is to use signal processing tools to analyze the obtained EEG signals, such as wavelet transforms, and classify them into different classes. The features of the EEG were extracted using wavelet transform and autoregressive models. After feature extrusion, the secondary goal is to improve the accuracy of the classification. To achieve this, we applied a neural network classifier based on reverse propulsion. Characteristic infarctions and classifications from 100 subjects in each group were analyzed and the data was divided into training, testing and verification of the proposed algorithm.

Published

2020-12-31

Issue

Section

Articles