An Effective Classification Method for BCI Based on Optimized SVM by GA

This paper proposed an effective method for EEG data classification in a Brain-Computer Interfacing system. We use Principal Component Analysis for feature extracting, then use an optimized Support Vector Machine for classification. The SVM's parameters are optimized by Genetic Algorithm. Furthermore, and optimal signal combination search is performed to get a higher classification rate, an explanation from the human physiological point of view is given. Experiment shows that this method can achieve higher classification accuracy than normal SVM classifier and artificial neural network.