The Research On Pattern Recognition And Classification Of Brain Function Based On SVM

EEG are bioelectrical activity generated by the central nervous system ,it contains a lot of the information about status and changes in the nervous system. It has become the forefront of current research that the correct interpretation of characteristics of EEG. It is a crucial step of better understanding of brain function, and knowing the link between peripheral nerve and brain.In this paper, we research brain function pattern recognition and classification by using support vector machine (SVM). Because the collected EEG signal includes noise, we must de-noise before classifying. In order to more accurately identify brain function pattern, we analyze and extract features from EEG signals.Details include:Firstly, this paper provides an overview of the development of Support Vector Machine Status, the history and the research situation of EEG study, and the situation of identification and classification of brain function pattern by using SVM.Secondly, this paper introduces the concept of EEG, and the major research methods dealing with EEG.Thirdly, details the statistical learning theory and support vector machine theoryLastly, in order to more efficiently extract the EEG wave, a new extraction method link with wavelet packet technology is studied. First of all we decompose the EEG with the method of wavelet packet, and reconstruct the EEG single in different frequency bands. Then analyze power spectrum and calculate the energy. Finally, train and test the extracted feature vector by using SVM, implement the EEG pattern recognition and classification, and achieved good classification results.