EEG signals combined with deep learning play an important role in the study of human–computer interaction. However, the limited dataset makes it challenging to study EEG signals using deep learning methods. Inspired by the GAN network in image generation, this paper presents an improved generative adversarial network model L-C-WGAN-GP to generate artificial EEG data to augment training sets and improve the application of BCI in various fields. The generator consists of a long short-term memory (LSTM) network and the discriminator consists of a convolutional neural network (CNN) which uses the gradient penalty-based Wasserstein distance as the loss function in model training. The model can learn the statistical features of EEG signals and generate EEG data that approximate real samples. In addition, the performance of the compressed sensing reconstruction model can be improved by using augmented datasets. Experiments show that, compared with the existing advanced data amplification techniques, the proposed model produces EEG signals closer to the real EEG signals as measured by RMSE, FD and WTD indicators. In addition, in the compressed reconstruction of EEG signals, adding the new data reduces the loss by about 15% compared with the original data, which greatly improves the reconstruction accuracy of the EEG signals’ compressed sensing.

Background/Objectives: Emotions can be recognized through external behavioral cues and internal physiological signals. Owing to the inherently complex psychological and physiological nature of emotions, models relying on a single modality often suffer from limited robustness. This study aims to improve emotion recognition performance by effectively integrating electroencephalogram (EEG) signals and facial expressions through a multimodal framework. Methods: We propose a multimodal emotion recognition model that employs a Multi-Scale Dilated Attention Convolution (MSDAC) network tailored for facial expression recognition, integrates an EEG emotion recognition method based on three-dimensional features, and adopts a self-learning decision-level fusion strategy. MSDAC incorporates Multi-Scale Dilated Convolutions and a Dual-Branch Attention (D-BA) module to capture discontinuous facial action units. For EEG processing, raw signals are converted into a multidimensional time–frequency–spatial representation to preserve temporal, spectral, and spatial information. To overcome the limitations of traditional stitching or fixed-weight fusion approaches, a self-learning weight fusion mechanism is introduced at the decision level to adaptively adjust modality contributions. Results: The facial analysis branch achieved average accuracies of 74.1% on FER2013, 99.69% on CK+, and 98.05% (valence)/96.15% (arousal) on DEAP. On the DEAP dataset, the complete multimodal model reached 98.66% accuracy for valence and 97.49% for arousal classification. Conclusions: The proposed framework enhances emotion recognition by improving facial feature extraction and enabling adaptive multimodal fusion, demonstrating the effectiveness of combining EEG and facial information for robust emotion analysis.