A GA-based feature selection of the EEG signals by classification evaluation: Application in BCI systems