Electroencephalography Analysis Using Neural Network and Support Vector Machine during Sleep

Abstract

The purpose of this paper is to analyze sleep stages accurately using fast and simple classifiers based on the frequency domain of electroencephalography(EEG) signal. To compare and evaluate system performance, the rules of Rechtschaffen and Kales(R&K rule) were used. Parameters were extracted from preprocessing process of EEG signal as feature vectors of each sleep stage analysis system through representatives of back propagation algorithm and support vector machine (SVM). As a result, SVM showed better performance as pattern recognition system for classification of sleep stages. It was found that easier analysis of sleep stage was possible using such simple system. Since accurate estimation of sleep state is possible through combination of algorithms, we could see the potential for the classifier to be used for sleep analysis system.