Research on the Classification of Brain Function Based on SVM

Abstract

In order to improve the accuracy in classification of electroencephalographic (EEG) signals of different brain functions, the research of how to select the suitable classifier is carried out in the paper. Some experiments have been performed to select the suitable kernel function of support vector machine (SVM). Four different kernel functions are put into the comparison and the results show that the radial basis function (RBF) has the highest rate of correct classification (RCC). On the other hand, the EEG signals from different electrodes will lead different classification results. The study of selecting the suitable electrode has been done. It shows that the RCC of the signal which from the electrode near by the area where EEG signal of a certain brain function is generated is much higher than those far from. We increase the dimension of SVM through combine the signal of different channels, which the RCC is very low, to improve the RCC of the signal which far from the area of the certain brain function. The results of our experiments are satisfied. The RCC of the EEG signal can reached to 99%.