Abstract

Info

Title:: Brain-computer interface of motion imagery based on mutual information-based feature extraction

Author(s):: JIA Tingting1; 2; DONG Chaoyi1; 2; MA Shuang1; 2; MA Pengfei1; 2; CHEN Xiaoyan1; 2; XIAO Zhiyun1; 2; QI Yongsheng1; 2; 1. College of Electric Power, Inner Mongolia University of Technology, Hohhot 010088, China 2. Inner Mongolia Key Laboratory of Electromechanical Control, Hohhot 010051, China

Keywords:: Keywords: mutual information motion imagery autoregressive model support vector machine brain-computer interface

Abstract:: Abstract: Brain-computer interface is a kind of system that realizes communication between computers and human brains and other devices. The features such as the network connection structure weights of multi-channel motion imagery clectroencephalogram (EEG) signals from F3, F4, C3, C4, FZ, CZ, FC1, FC2, FC5, FC6, etc are introduced in the study, and support vector machine is used to classify different motion imagery tasks. The proposed brain network structure feature extraction based on mutual information (MI) is compared with the traditional autoregressive model-based parameter feature extraction. It is found that the accuracy rate of the MI-based feature extraction for motion imagery EEG signals classification is significantly higher than that of the autoregressive model-based parameter feature extraction. After the two types of features are fused, the classification accuracy rate of the constructed brain-computer interface classifier is significantly higher than that of feature extraction method using MI or autoregressive model alone.