/Emotion-Recognition-based-on-EEG-using-Generative-Adversarial-Nets-and-Convolutional-Neural-Network

Emotion recognition plays an important role in the field of human-computer interaction (HCI). Automatic emotion recognition based on EEG is an important topic in brain-computer interface (BCI) applications.Currently, deep learning has been widely used in the field of EEG emotion recognition and has achieved remarkable results. However, due to the small amount of EEG data and the serious imbalance in the proportion of EEG data categories, it is difficult to use deeper models. In addition, we believe that there is a frequency band correlation feature between the EEG signal frequency bands, which has an important effect on EEG emotion recognition. In this paper, we first proposed an adversarial neural network model for sample generation. Because we used PSD features in the experiment, this generative model is called PSD-GAN. Then we designed FBSCNN(Frequency band separation convolutional neural network) and FBCCNN(Frequency Band Correlation Convolutional Neural Network) models as a comparison to explore the influence of frequency band correlation features on EEG emotion recognition. Among them, FBSCNN can not extract the frequency band correlation features, but FBCCNN can extract the frequency band correlation features. The experimental results show that the samples generated by PSD-GAN have good performance, and the frequency band correlation feature can effectively improve the accuracy of EEG emotion recognition. Moreover, we compare our FBCCNN + PSD-GAN model with similar studies and the results show that our model is highly competitive.

This repository is not active