An EEG-Based Human Emotion Recognition Model Using Attention Hierarchical Variational EEGNet

Abstract

The physiological factors and decisions we make daily are greatly affected by a person's emotional state. An appropriate subfield of study and research is how brain–computer interface devices can be used to identify people's emotional states. The electroencephalogram (EEG) is used to understand why there are changes over time in our physiological signal patterns that help us identify how we feel about something. The EEG signal is used more often now in combination with machine learning and deep learning algorithms to assist in determining a person's emotional state. The model used in this article is called the attention-based hierarchical variational EEGNet. By using state-of-the-art learning techniques to process noisy and changing EEG signals, the proposed method is expected to be better than traditional, custom, and shallow learning. The performance of this new model was evaluated using the SJTU Emotion EEG Dataset. According to the results of the study, the proposed method for classifying people's emotional states is 96.5% accurate, and therefore shows competitive performance than typical state-of-the art methods.