Improved feature extraction and classification methods for electroencephalographic signal based brain-computer interfaces

摘要

It is very important for brain-computer interfaces to classify accurately electroencephalographic signals. In this paper, we proposed an improved feature extraction and classification methods for electroencephalographic signal-based brain-computer interfaces. We decompose electroencephalogram signal into bands using discrete wavelet transform and compute the approximate entropy values. The feature vectors are selected adaptively from statistical wavelet coefficients and approximate entropy values. The support vector machine is used to classify the features. The experimental results demonstrate the proposed system has great performance and reliability.