With the continuous development of portable noninvasive human sensor technologies such as brain–computer interfaces (BCI), multimodal emotion recognition has attracted increasing attention in the area of affective computing. This paper primarily discusses the progress of research into multimodal emotion recognition based on BCI and reviews three types of multimodal affective BCI (aBCI): aBCI based on a combination of behavior and brain signals, aBCI based on various hybrid neurophysiology modalities and aBCI based on heterogeneous sensory stimuli. For each type of aBCI, we further review several representative multimodal aBCI systems, including their design principles, paradigms, algorithms, experimental results and corresponding advantages. Finally, we identify several important issues and research directions for multimodal emotion recognition based on BCI.

Domain adaptation (DA) tackles the problem where data from the source domain and target domain have different underlying distributions. In cross-domain (cross-subject or cross-dataset) emotion recognition based on EEG signals, traditional classification methods lack domain adaptation capabilities and have low performance. To address this problem, we proposed a novel domain adaptation strategy called adversarial discriminative-temporal convolutional networks (AD-TCNs) in this study, which can ensure the invariance of the representation of feature graphs in different domains and fill in the differences between different domains. For EEG data with specific temporal attributes, the temporal model TCN is used as the feature encoder. In the cross-subject experiment, our AD-TCN method achieved the highest accuracies of the valence and arousal dimensions in both the DREAMER and DEAP datasets. In the cross-dataset experiment, two of the eight task groups showed accuracies of 62.65% and 62.36%. Compared with the state-of-the-art performance in the same protocol, experimental results demonstrated that our method is an effective extension to realize EEG-based cross-domain emotion recognition.

In this paper, a novel spatio-temporal self-constructing graph neural network (ST-SCGNN) is proposed for cross-subject emotion recognition and consciousness detection. For spatio-temporal feature generation, activation and connection pattern features are first extracted and then combined to leverage their complementary emotion-related information. Next, a self-constructing graph neural network with a spatio-temporal model is presented. Specifically, the graph structure of the neural network is dynamically updated by the self-constructing module of the input signal. Experiments based on the SEED and SEED-IV datasets showed that the model achieved average accuracies of 85.90% and 76.37%, respectively. Both values exceed the state-of-the-art metrics with the same protocol. In clinical besides, patients with disorders of consciousness (DOC) suffer severe brain injuries, and sufficient training data for EEG-based emotion recognition cannot be collected. Our proposed ST-SCGNN method for cross-subject emotion recognition was first attempted in training in ten healthy subjects and testing in eight patients with DOC. We found that two patients obtained accuracies significantly higher than chance level and showed similar neural patterns with healthy subjects. Covert consciousness and emotion-related abilities were thus demonstrated in these two patients. Our proposed ST-SCGNN for cross-subject emotion recognition could be a promising tool for consciousness detection in DOC patients.

Electroencephalogram (EEG) signals have been widely used in emotion recognition. However, the current EEG-based emotion recognition has low accuracy of emotion classification, and its real-time application is limited. In order to address these issues, in this paper, we proposed an improved feature selection algorithm to recognize subjects’ emotion states based on EEG signal, and combined this feature selection method to design an online emotion recognition brain-computer interface (BCI) system. Specifically, first, different dimensional features from the time-domain, frequency domain, and time-frequency domain were extracted. Then, a modified particle swarm optimization (PSO) method with multi-stage linearly-decreasing inertia weight (MLDW) was purposed for feature selection. The MLDW algorithm can be used to easily refine the process of decreasing the inertia weight. Finally, the emotion types were classified by the support vector machine classifier. We extracted different features from the EEG data in the DEAP data set collected by 32 subjects to perform two offline experiments. Our results showed that the average accuracy of four-class emotion recognition reached 76.67%. Compared with the latest benchmark, our proposed MLDW-PSO feature selection improves the accuracy of EEG-based emotion recognition. To further validate the efficiency of the MLDW-PSO feature selection method, we developed an online two-class emotion recognition system evoked by Chinese videos, which achieved good performance for 10 healthy subjects with an average accuracy of 89.5%. The effectiveness of our method was thus demonstrated.