EEG Motor Imagery Signal Classification Using Firefly Support Vector Machine

This Study proposed Firefly-Support vector machine (FASVM) method to improve the accuracy of EEG motor imagery signals classification. Redundancy of features in EEG signal classification can affect its performance. Firefly is a metaheuristic optimization based on firefly behavior. Firefly algorithm is used to select optimal subset of features in order to increase the classification accuracy. Common Spatial Pattern (CSP) is used to extract EEG signals features before selected by firefly algorithm. The purpose of CSP is to maximize the variance for one class and to minimize the other class variance. This study used BCI Competition III data set IVa. Feature vector extracted from CSP is selected using FASVM. The accuracy of SVM is used as objective function for firefly algorithm optimisation. The proposed method FASVM produced good result with average accuracy of 93.20 %.