Self-Supervised EEG Emotion Recognition Models Based on CNN

Abstract

Emotion plays crucial roles in human life. Recently, emotion classification from electroencephalogram (EEG) signal has attracted attention by researchers due to the rapid development of brain computer interface (BCI) techniques and machine learning algorithms. However, recent studies on emotion classification show resource utilization because they use the fully-supervised learning methods. Therefore, in this study, we applied the self-supervised learning methods to improve the efficiency of resources usage. We employed a self-supervised approach to train deep multi-task convolutional neural network (CNN) for EEG-based emotion classification. First, six signal transformations were performed on unlabeled EEG data to construct the pretext task. Second, a multi-task CNN was used to perform signal transformation recognition on the transformed signals together with the original signals. After the signal transformation recognition network was trained, the convolutional layer network was frozen and the fully connected layer was reconstructed as emotion recognition network. Finally, the EEG data with affective labels were used to train the emotion recognition network to clarify the emotion. In this paper, we conduct extensive experiments from the data scaling perspective using the SEED, DEAP affective dataset. Results showed that the self-supervised learning methods can learn the internal representation of data and save computation time compared to the fully-supervised learning methods. In conclusion, our study suggests that the self-supervised machine learning model can improve the performance of emotion classification compared to the conventional fully supervised model.

Self-supervised deep learning and EEG categorization

“…Self-supervised strategies are mostly used to pretrain DNNs from which classifiers are created and many show that the performance of supervised learning can be increased (Daoud et al, 2020;Eldele et al, 2023;Hermans et al, 2023;Li et al, 2020;Kostas et al, 2021;Luo et al, 2022;Ou et al, 2022;Qui et al, 2018;Sahani et al, 2021;Sartipi et al, 2024;Vařeka et al, 2017;Wang et al, 2023B;Wulsin et al, 2011;Yang et al, 2023;Yuan et al, 2019). There are also examples of training DNNs self-supervised and using their output as features for non-deep machine learning classification (Dairi et al, 2022;Liu et al, 2023, Hassan Shah et al, 2023Supratak et al, 2014;Tautan et al, 2019, Zhou et al, 2019.…”

Research on the Strong Generalization of Coal Gangue Recognition Technology Based on the Image and Convolutional Neural Network under Complex Conditions

“…Since 2012, the convolutional neural network has made major breakthroughs in the field of image recognition, and its recognition capabilities have surpassed humans . Therefore, the convolutional neural network is also widely used in emotion classification, , speech recognition, , agricultural engineering, , defect detection, fault diagnosis, and other fields and has made significant breakthroughs and improvements. Taking coal gangue recognition as an example, the convolutional neural network has been widely used and has achieved important research results.…”

Self-supervised motor imagery EEG recognition model based on 1-D MTCNN-LSTM network

“…Nevertheless, recruiting a large number of subjects and collecting and labeling EEG data is a challenging and time-consuming task. Therefore, the sample size of EEG data is usually limited [19].…”