Classification of EEG in A Steady State Visual Evoked Potential Based Brain Computer Interface Experiment

Abstract

In this paper, electroencephalogram (EEG) signals of 20 subjects are classified in a steady state visual evoked potential (SSVEP) based brain computer interface (BCI) system by using 4 different stimulation frequencies in a program created by Visual C#. After applying proper pre-processing methods, power spectral density (PSD) based features are extracted around first and second harmonics of the stimulation frequencies. Average classification performance obtained from 20 subjects in 4-class classification is 83.62% with Nearest Mean Classifier (NMC). Results for 5-class classification, EEG segment size and gender differences are also analyzed in a detailed manner. The classification method is simple and very suitable for real-time experiments.