Classification of EEG signals using time domain features

Abstract

Electroencephalogram (EEG) signals are widely used in many fields such as clinical diagnosis, Brain-Computer Interface, performance measurement and emotion analysis. The most important benefit of EEG signal analysis is to control a device without moving muscles or provide communication. In particular, patients with ALS (Amyotrophic Lateral Sclerosis) disease who cannot control muscles can communicate with their environment. Researchers use signal processing and machine learning in order to extract meaningful information from raw EEG signals. In this study, data obtained at the University of Tubingen in Germany and presented in 2003 BCI competition is classified using only time domain features and nonlinear classifier. Classification accuracy of time domain features without preprocessing is higher than that of the accuracy of BCI competition.