“…This might be because there is the significant difference of static EEG signals among different time periods, which has a great interference to classification. [20], support vector machine using DE features (SVM) [20], minimalist neural network (MNN) [26], space-temporal recurrent neural network (STRNN) [14], dynamical graph convolutional neural networks (DGCNN) [18], Bimodal deep autoencoder (BDAE) [22], Graph regularized extreme learning machine (GELM) [27], and SyncNet [28]. We apply several representative methods on our dataset for comparison, the methods and models involved are as follows: PSD+SVM: Classical emotion recognition method on EEG signals, where PSD features are extracted from each channel of EEG signals at five specific frequency bands (delta: 1-3 Hz, theta: 4-7 Hz, alpha: 8-13 Hz, beta: 14-30 Hz, gamma: 31-50 Hz), and are fed into the traditional SVM.…”