In the BCI system, translation is needed to transform the
detected feature of a subject into a command. The
translation is done through a classification algorithm. A
classification algorithm uses distinctive features in
identifying the class to which a test signal belongs. Widely
used classification methods in the EEG based BCI field,
have been adopted from the pattern recognition community,
including the linear discriminant analysis (LDA), the
support vector machine (SVM), and the k-nearest-neighbor
(kNN) [5].