Due to the intracranial volume conduction effects, high-dimensional multi-channel electroencephalography (EEG) features often contain substantial redundant and irrelevant information. This issue not only hinders the extraction of discriminative emotional representations but also compromises real-time performance. Feature selection has been established as an effective approach to address the challenges while enhancing the transparency and interpretability of emotion recognition models. However, existing EEG feature selection research overlooks the influence of latent EEG feature structures on emotional label correlations and assumes uniform importance across various channels, directly limiting the precise construction of EEG feature selection models for multi-dimensional affective computing. To overcome these issues, this paper proposes a channel-wise EEG feature selection (CWEFS) method for multi-dimensional emotion recognition. Inspired by the volume conduction effects, CWEFS models feature selection within a shared latent structure that captures a consensus representation across EEG channels. This consensus space is jointly learned with a latent semantic analysis of emotional labels to preserve local geometric structure. Furthermore, CWEFS incorporates adaptive channel-weight learning to automatically assess the contribution of each channel. Comprehensive experimental results, compared against nineteen popular feature selection methods, demonstrate that the EEG feature subsets chosen by CWEFS achieve optimal emotion recognition performance across six evaluation metrics.

A major challenge in motor imagery Brain-Computer Interfaces (MI-BCIs) arises from domain shift due to large individual differences. Currently, most cross-subject MI-BCI decoding methods rely on transfer learning to extract subject-shared features or align data distributions. However, these methods typically require all unlabeled data from the target subjects or labeled calibration data, which is unavailable in practical applications. To address this, we propose a brain lateralization-guided subject adaptive network, BLSAN, to enhance model generalization through local-global adversarial training. Specifically, two separate adversarial networks for left and right hemispheres are designed to reduce local differences, and features extracted from both hemispheres are combined for global adversarial training. Additionally, we design a confidence-based pseudo label generation method to enhance model discriminability. We validate the effectiveness of our approach on two public MI datasets, BCI Competition IV 2a and 2b, only with some unlabeled calibration data, which improves the practicality of MI-BCIs.

Cross-site datasets expand the data size and could improve the disease diagnosis capabilities of machine learning models. However, differences in data distribution between different sites can lead to poor model generalizability. Although transfer learning is a mainstream method often used to tackle the issue, most transfer learning studies assume that all the target samples are given in the training procedure, which is not available in clinical applications where the target samples arrive sequentially. Online transfer learning (OTL) aims to accomplish clinical diagnostic tasks by adaptively updating the ensemble model containing source and target classifiers. However, OTL is limited by zero initialization and requires numbers of samples to iterate. This results in the underperformance of OTL in clinical applications where few samples can be obtained. In this article, we propose a new framework named few-shot synthetic OTL (FSOTL) to address this issue. FSOTL uses synthetic data to warm up the model in an online fashion. It not only alleviates the problem of scarcity of samples in the target domain but also enables the model to gain more knowledge. Our experiments show that FSOTL performs more stably and achieves more accurate results with few target samples, thereby offering a promising cross-site online computer-aided diagnosis system for large-scale applications.

Although advances in deep learning technologies have greatly facilitated the brain intention decoding from electroencephalogram (EEG) in motor imagery brain–computer interfaces (MI-BCIs), significant individual differences hinder the practical cross-subject MI-BCI applications. Unlike other existing domain adversarial transfer networks that focus on designing different discriminators to reduce individual differences, inspired by the motor lateralization phenomenon, we innovatively utilize transformer and the spatiotemporal pattern differences of EEG as prior knowledge to enhance the feature discriminability in our brain decoding adversarial network. In addition, to address adversarial network decision boundaries bias toward the source domain, we propose a data augmentation method, EEGMix to rapidly mix and enrich the target domain data. With an adaptive adversarial factor, our decoding model reduces the differences in marginal and conditional distribution simultaneously. Three public MI datasets, 2a, 2b, and OpenBMI verified our model's effectiveness. The accuracy achieved 77.49%, 85.19%, and 79.37%, superior to other state-of-the-art algorithms.

A common drawback of EEG-based emotion recognition is that volume conduction effects of the human head introduce interchannel dependence and result in highly correlated information among most EEG features. These highly correlated EEG features cannot provide extra useful information, and they actually reduce the performance of emotion recognition. However, the existing feature selection methods, commonly used to remove redundant EEG features for emotion recognition, ignore the correlation between the EEG features or utilize a greedy strategy to evaluate the interdependence, which leads to the algorithms retaining the correlated and redundant features with similar feature scores in the EEG feature subset. To solve this problem, we propose a novel EEG feature selection method for emotion recognition, termed global redundancy minimization in orthogonal regression (GRMOR). GRMOR can effectively evaluate the dependence among all EEG features from a global view and then select a discriminative and nonredundant EEG feature subset for emotion recognition. To verify the performance of GRMOR, we utilized three EEG emotional data sets (DEAP, SEED, and HDED) with different numbers of channels (32, 62, and 128). The experimental results demonstrate that GRMOR is a promising tool for redundant feature removal and informative feature selection from highly correlated EEG features.

Feature selection has attracted a lot of attention in obtaining discriminative and non-redundant features from high-dimension data. Compared with traditional filter and wrapper methods, embedded methods can obtain a more informative feature subset by fully considering the importance of features in the classification tasks. However, the existing embedded methods emphasize the above importance of features and mostly ignore the correlation between the features, which leads to retain the correlated and redundant features with similar scores in the feature subset. To solve the problem, we propose a novel supervised embedded feature selection framework, called feature selection under global redundancy minimization in orthogonal regression (GRMOR). The proposed framework can effectively recognize redundant features from a global view of redundancy among the features. We also incorporate the large margin constraint into GRMOR for robust multi-class classification. Compared with the traditional embedded methods based on least square regression, the proposed framework utilizes orthogonal regression to preserve more discriminative information in the subspace, which can help accurately rank the importance of features in the classification tasks. Experimental results on twelve public datasets demonstrate that the proposed framework can obtain superior classification performance and redundancy removal performance than twelve other feature selection methods.

A common drawback of the EEG applications is that the volume conduction of human head leads to lots of redundant information in EEG recordings. To reduce the redundancy and choose informative EEG features, in this paper, we propose an EEG feature selection technique, termed as Feature Selection with Orthogonal Regression (FSOR). Compared with classical feature selection methods, for nonlinear and nonstationary EEG signals, FSOR can employ orthogonal regression to preserve more discriminative information in the subspace. To verify the EEG feature selection performance, we collected a multichannel EEG dataset for emotion recognition and compared FSOR with two popular feature selection methods. The experimental results demonstrate the advantage of FSOR method over others for reducing the redundant information among the EEG relevant features. Additionally, we found that the absolute power ratio of beta wave to theta wave is the most discriminative feature, and beta band is the critical band for emotion recognition.