SECNN-LSTM: A CNN and efficient Attention Mechanism Based LSTM network for Electroencephalogram (EEG)-Based Emotion Recognition

Electroencephalogram (EEG)-based emotion recognition has encountered challenges that limit generalization challenges, these challenges limit its effectiveness in real-world applications. To address this problem, we propose a SECNN-LSTM network, it uses multi-channel EEG to learn more distinguished emotional representations automatically. First, we preprocess raw EEG signals by using filtering and noise reduction techniques. Second, to extract EEG spatial feature information mainly, we use the CNN combined with attention. Finally, by utilizing the modeling ability of LSTM network on sequences, the emotional features of 32 channels are combined into a feature sequence, combined with attention, multi-channel fused emotional features are extracted. Our proposed method is evaluated using public DEAP datasets, and the experimental results demonstrate its superior classification accuracy compared to other well-known methods.