Emotion Recognition of EEG Based on Dual-Input Multi-Network Fusion Features

Aiming at the problems of insufficient feature extraction and low emotion recognition accuracy caused by the existing EEG emotion recognition network structure, a multi-branch network EEG emotion recognition method is proposed, and a dual-input and three-branch network (DTNET) model is constructed. The model takes EEG time series samples and 3D spatiotemporal matrix as inputs, and features of EEG signals are extracted by combining convolution neural network, bidirectional gated loop unit and echo state network. Experimental results show that the model has high performance.