Comparison between Alcoholic and Control Subjects in EEG signals Using Classification Methods

Abstract

Alcoholism could be identified through analyzing electroencephalogram (EEG) signals. Yet, it is difficult to analyze with multi-channel EEG signal since it is frequently needing long time for execution and complex calculations. The presented paper proposed 13 optimal channel to feature extraction. Firstly, 1200 recordings of biomedical signals will be presented for extracting the sample entropy. Statistical analysis approach will be utilized for the purpose of choosing the best channels for identifying abnormalities in alcoholics. Secondly four classifiers are applied at the decision level, Naive Bayes, SVM, Logistic Regression, KNN, the accuracy was 80.1%,92.5%, 73.7% and 90.3%Respectively, in this study the SVM classifier is more accuracy.