Abstract

The Brain-Computer Interface (BCI) is a system that enables individuals who cannot use the existing muscle and nervous system because of various reasons to communicate with the environment. Steady-state visual evoked potentials (SSVEP) from EEG signals have gained wide research interest due to their high signal-to-noise ratio and higher information transfer rate compared to other BCI techniques. Therefore SSVEP plays a major role in practical applications. In this study, the data set (AVI SSVEP Dataset) obtained through open access from the Internet (www.setzner.com) was analyzed. In the dataset, electroencephalography (EEG) signals were recorded in which the participants were looking at a flickering box at eight distinct frequencies (6, 6.5, 7, 7.5, 8.2, 9.3, 10 and 12 Hz) whose color changes rapidly from black to white. We extracted twenty-five features containing only time-domain properties from SSVEP signals to predict which frequency was applied to the subject. These features were applied to classifiers of Decision Tree, Discriminant Analysis, Naive Bayes, Support Vector Machines, k-Nearest Neighbors, and Ensemble Classifiers. We obtained the maximum accuracy of 42.9% for each subject separately. When we evaluate all subjects using the same classifier, we achieved a 20% accuracy. K-Nearest Neighbors and Ensemble Classifier give the best classification performance in all experiments in this study.