Feature extraction of brain-computer interface based on improved multivariate adaptive autoregressive models

Feature extraction of EEG signals plays an important role for classifying spontaneous mental activities in EEG-based brain computer interface (BCI). For the non-stationary nature of EEG data makes necessary some kind of adaptation of the BCI system, an improved feature extraction method based on multivariate adaptive autoregressive (MVAAR) models is proposed and applied to the classification of Motor imagery. In this paper, three subjects participated in the BCI experiment which contains three mental tasks including imagination of left hand, right hand and foot movement. After preprocessing, improved MVAAR was applied to extract the feature of EEG signals. Then, Linear Discriminant Analysis (LDA) was used to classify the feature extracted. After that, a comparison of feature extract methods between MVAAR and other methods was made. The result shows that MVAAR is an effective feature extraction method especially for online BCI system.