Self-supervised Group Meiosis Contrastive Learning for EEG-Based Emotion Recognition

Abstract

The progress of EEG-based emotion recognition has received widespread attention from the fields of human-machine interactions and cognitive science in recent years. However, how to recognize emotions with limited labels has become a new research and application bottleneck. To address the issue, this paper proposes a Self-supervised Group Meiosis Contrastive learning framework (SGMC) based on the stimuli consistent EEG signals in human being. In the SGMC, a novel geneticsinspired data augmentation method, named Meiosis, is developed. It takes advantage of the alignment of stimuli among the EEG samples in a group for generating augmented groups by pairing, cross exchanging, and separating. And the model adopts a group projector to extract group-level feature representations from group EEG samples triggered by the same emotion video stimuli. Then contrastive learning is employed to maximize the similarity of group-level representations of augmented groups with the same stimuli. The SGMC achieves the state-of-the-art emotion recognition results on the publicly available DEAP dataset with an accuracy of 94.72% and 95.68% in valence and arousal dimensions, and also reaches competitive performance on the public SEED dataset with an accuracy of 94.04%. It is worthy of noting that the SGMC shows significant performance even when using limited labels. Moreover, the results of feature visualization suggest that the model might have learned video-level emotionrelated feature representations to improve emotion recognition. And the effects of group size are further evaluated in the hyper parametric analysis. Finally, a control experiment and ablation study are carried out to examine the rationality of architecture. The code is provided publicly online 1 .

Applications of Self-Supervised Learning to Biomedical Signals: where are we now

“…SimCLR, BYOL, CPC) already proposed for general purpose tasks and introduce some domain knowledge modifications to better suit the investigated medical task. Some works proposed more biological inspired data augmentation techniques for the generation of the positive/negative pairs [113]. Others focused on the way similarity between pairs is evaluated, for example by modifying the objective learning function.…”

“…Zhang et al [117] proposed GANSER, a generative self-supervised framework which combines an adversarial augmentation network (AAN) and a multi-factor training network (MTN). Finally, Shen et al [115] and Kan et al [113] proposed two novel contrastive learning approaches. The first, Contrastive learning for Inter-Subject Alignment (CLISA), tries to maximize the similarity in EEG signal representations across subjects who received the same (emotional) stimuli.…”

Applications of Self-Supervised Learning to Biomedical Signals: A Survey

“…[143] and Kan et al . [141] proposed two novel contrastive learning approaches. The first, Contrastive Learning for Inter-Subject Alignment (CLISA), tries to maximize the similarity in EEG signal representations across subjects who received the same emotional stimuli, hence without resorting to standard data augmentation procedures.…”

A study on the application of contrastive learning in the brain-computer interface of motor imagery

“…To achieve the main goal of learning universal feature representation, which allows the use of these universal representations in multiple tasks in the fine-tuning stage, multiple forms of pre-tasks and corresponding objective functions need to be used to learn feature representation for different forms of tasks. Compared with supervised learning that completes training using labeled data, self-supervised learning can extract internal features of data, making them suitable for downstream tasks, thereby improving the performance of downstream tasks [6] . Through self-supervised learning, largescale unlabeled data can be more effectively utilized to improve the performance of models, which cannot be achieved by supervised learning.…”