A Model for EEG-Based Emotion Recognition: CNN-Bi-LSTM with Attention Mechanism

• DOI: 10.3390/electronics12143188
• Publication Date: 2023-07-24T05:12:28Z
• Authors: Zhentao Huang; Yahong Ma; Rongrong Wang; Weisu Li; Yongsheng Dai
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology; convolutional neural network (CNN); electroencephalograph (EEG); bi-directional long short-term memory (Bi-LSTM); attention mechanism; emotion signal recognition

Emotion analysis is the key technology in human–computer emotional interaction and has gradually become a research hotspot field of artificial intelligence. The problems emotion based on EEG are feature extraction classifier design. existing methods mainly use machine learning rely manually extracted features. As an end-to-end method, deep can automatically extract features classify them. However, most models recognition still need manual screening data pre-processing, accuracy convenience not high enough. Therefore, this paper proposes CNN-Bi-LSTM-Attention model to emotions signals. original used as input, CNN Bi-LSTM network for fusion, then electrode channel weights balanced through attention mechanism layer. Finally, signals classified different kinds emotions. An classification experiment conducted SEED dataset evaluate performance proposed model. experimental results show that method effectively was assessed two distinctive tasks, one with three four target classes. average ten-fold cross-validation 99.55% 99.79%, respectively, corresponding which significantly better than other methods. It be concluded our superior recognition, widely many fields, including modern neuroscience, psychology, neural engineering, computer science well.