Research on two-dimensional convolution-based EEG emotion recognition methods

In our daily lives, emotions affect our behavior, language, thoughts, etc. from time to time. Emotions include people's subjective experiences, physiological reactions, and behavioral responses. As people pay more attention to their daily lives, emotions gradually play an important role in them. In this paper, a two-dimensional convolution-based emotion recognition method is proposed. The frequency and time signals of EEG are used as inputs in a multi-channel format to perform three levels of emotion classification tasks. The proposed method is finally applied to the SEED dataset using the proposed method. The experimental results show that compared to other classification algorithms, the performance of 2D-CNN in terms of accuracy in emotion recognition is superior, laying the foundation for subsequent research in the field of human-computer interaction.