EEG-Based Emotion Recognition Using Convolutional Neural Network with Functional Connections

Emotion recognition plays a vital role in Brain-Computer Interaction. To extract and employ the inherent information implied by functional connections among EEG electrodes, we propose a multichannel EEG emotion recognition method using convolutional neural network (CNN) with functional connectivity as input. Specifically, the phase synchronization indices are employed to compute the EEG functional connectivity matrices. Then a CNN is proposed to effectively extract the classification information of these functional connections. The experimental results based on the DEAP and SEED datasets validate the superior performance of the proposed method, compared with the input of raw EEG data. The code of the proposed model is available at https://github.com/deep-bci/ERBCPSI.