Abstract:

The feature extraction was performed from the electroencephalography signals in the brain-computer interface for two different mental tasks(imagine to move the left land or right hand) using the wavelet entropy and the band power as the combining feature. The Fisher linear discrimination analysis was used to classify the features. The classification accuracy and the mutual information were used as evaluation criteria. The results of off-line analysis showed that the proposed algorithm is characterized by good classification accuracy and mutual information, providing a new way for the feature extraction and classification of the mental tasks in the braincomputer interface.

Key words: information processing, braincomputer interface, wavelet entropy, linear discrimination analysis, mutual information, band power