Improving classification accuracy using fuzzy method for BCI signals

Electroencephalograph (EEG) signals feature extraction and processing is one of the most difficult and important part in the brain-computer interface (BCI) research field. EEG signals are generally unstable, complex and have low signal-noise ratio, which is difficult to be analyzed and processed. To solve this problem, this paper disassembles EEG signals with the empirical mode decomposition (EMD) algorithm, extracts the characteristic values of the major intrinsic mode function (IMF) components, and then classifies them with fuzzy C-means (FCM) method. Also, comparison research is done between the proposed method and several current EEG classification methods. Experimental results show that the classification accuracy based on the EEG signals of the second BCI competition in 2003 is up to 78%, which is superior to those of the comparative EEG classification methods.