Classification of Mental Task EEG Signals Using Wavelet Packet Entropy and SVM

This paper address on the classification of mental task EEG signals, which is one of the key issues of Brain-Computer Interface (BCI). We proposed a method using wavelet packet entropy and Support Vector Machine (SVM). First, we apply 7 levels wavelet packet decomposition to each channel of EEG with db4. After extraction four spectrum bands (δ,θ,α,β), an entropy algorithm was performed on each bands. The resulting entropy vectors are then used as inputs to SVM to train and test. We test the method on EEG signals during 5 mental tasks collected by 2 subjects. The accuracy on 2-class classification for subject 1 is averaged 93.0%, and 87.5% for subject 2. The results also show that our method outperforms the classical methods for multi-class problems.