An Empirical Analysis of Different Machine Learning Techniques for Classification of EEG Signal to Detect Epileptic Seizure

Abstract

Electroencephalogram (EEG) signal is a modest measure of electric flow in a human brain. It is responsible for information flow through the neurons in the brain which controls and monitors the full torso. Hence, to wide and in-depth understand of their effectiveness in EEG signal analysis is the primary focus of this paper. Moreover, machine learning techniques often proven as more efficacious compared to other techniques. To this effect, the present study primarily focuses on the analysis of EEG signal through the classification of the processed data by discrete wavelet transform (DWT) for identification of epileptic seizures using machine learning techniques. Machine learning techniques like neural networks and support vector machine (SVM) are the focus of this paper for classification of EEG signals to label epilepsy patients. In neural networks, the empirical analysis gives focus on multi-layer perceptron, probabilistic neural network, radial basis function neural networks, and recurrent neural networks. Further, for multi-layer neural networks different propagation training algorithms are examined such as Back-Propagation, Resilient-Propagation, and Quick-Propagation. For SVM, several kernel methods are considered such as Linear, Polynomial, and RBF for empirical analysis. The analysis confirms with the present setting that, recurrent neural network is performed poor in all cases of prepared epilepsy data. However, SVM and probabilistic neural networks are very effective and competitive.