Study on classification algorithm of motor imagination EEG signal

Abstract

Motor imagination is an important area of brain-computer interface. In recent years, the application of deep learning algorithms has further improved the recognition rate of motor imagination EEG classification. However, the current deep learn-based motor imagination EEG studies mostly analyze the EEG as a matrix, ignoring the correlation between the electrode nodes that extract the EEG. Therefore, this paper attempts to propose a GCN-BILSTM model, which uses graph convolutional neural network to extract spatial features from EEG, and bidirectional long and short-term memory network to extract temporal features from EEG. This scheme has some advantages, because it requires less weight parameters and converges faster. In order to verify the superiority of the algorithm, the BCI-IV Dataset 2A is used to verify the algorithm proposed in this paper. Experiments show that the proposed algorithm can improve the recognition and classification accuracy of motor imagination EEG signals, and the classification accuracy of nine subjects reaches 84%, which verifies the effectiveness of the algorithm.