Convolutional recurrent neural network-based EEG signal classification in motor imagery

Brain-computer interface(BCI) is a method of extracting EEG signals by specific means and decoding them using signal processing algorithms to help people with motor disabilities to interact with the outside world through external devices. In order to improve the EEG signal pattern recognition rate of motor imagery. A pattern recognition method of neural network feature fusion combining convolutional neural network (CNN) and recurrent neural network (RNN) serial connections is proposed, and two different RNNs are used for experimental comparison. The proposed method is validated using the BCI Competition IV 2a dataset, and the experimental results show that the proposed method can effectively improve the multi-classification accuracy.