EEG Emotion Recognition based on Multi scale Self Attention Convolutional Networks

A multi-view self-attention module is proposed and paired with a multi-scale convolutional model to build a multi-view self-attention convolutional network for multi-channel EEG emotion recognition. First, time and frequency domain characteristics are extracted from multi-channel EEG signals, and a three-dimensional feature matrix is built using spatial mapping connections. Then, a multi-scale convolutional network extracts the high-level abstract features from the feature matrix, and a multi-view self-attention network strengthens the features. Finally, use the multilayer perceptron for sentiment classification. The experimental results reveal that the multi-view self-attention convolutional network can effectively integrate the time domain, frequency domain, and spatial domain elements of EEG signals using the DEAP public emotion dataset. The multi-view self-attention module can eliminate superfluous data, apply attention weight to the network to hasten network convergence, and enhance model recognition precision.