Research of Classification Methods and Evaluation Criteria in Brain-Computer Interface System

Brain-computer interface (BCI) system uses brain activity to control external devices such as computers and electronic devices. It is a novel kind of human computer interaction. BCI system can be regard as pattern recognition system, and the key point is classification of Electroencephalogram (EEG) signals under different mental tasks. Classification algorithms of BCI system include Fisher linear discriminant analysis, artificial neural network (ANN), and support vector machine (SVM). Evaluation criteria of BCI include accuracy rate, information transfer rate (ITR), and mutual information (MI). Basic theory, classification methods and evaluation criteria in BCI research are introduced in detail in this paper, and give much help to the development and application of BCI system.