GMSS: Graph-Based Multi-Task Self-Supervised Learning for EEG Emotion Recognition

Li, Yang; Chen, Ji; Li, Fu; Fu, Boxun; Wu, Hao; Youshuo, Ji,; Zhou, Yifeng; Niu, Yi; Shi, Guangming; Zheng, Wenming

doi:10.1109/taffc.2022.3170428

Cited by 28 publications

(20 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Li et al introduced a graph-based multi-task, self-supervised learning model that reduced the chance of overfitting for emotion recognition tasks. Their classification accuracy reached 89.18%, which is 5.74% lower than ours (Li et al 2022). Through a large number of experiments on the DEAP and SEED datasets, we can thus draw the conclusion that the ERGL method is effective.…”

Section: Experimental Results On Seedcontrasting

confidence: 51%

Spatial–temporal features-based EEG emotion recognition using graph convolution network and long short-term memory

Zheng

et al. 2023

Physiol. Meas.

View full text Add to dashboard Cite

Objective: emotion recognition on the basis of electroencephalography (EEG) signals has received a significant amount of attention in the areas of cognitive science and human-computer interaction (HCI). However, most existing studies either focus on one-dimensional EEG data ignoring the relationship between channels, or only extract time-frequency features while not involving spatial features. Approach: we develop spatial-temporal features based EEG emotion recognition using graph convolution network (GCN) and long short-term memory (LSTM) named ERGL. Firstly, the one dimensional EEG vector is converted into a two-dimensional mesh matrix, so that the matrix configuration corresponds to the distribution of brain regions at EEG electrode locations, thus to represent the spatial correlation between multiple adjacent channels in a better way. Secondly, the GCN and LSTM are employed together to extract spatial-temporal features, and GCN is used to extract spatial features, while LSTM units are applied to extract temporal features. Finally, the softmax layer is applied to emotion classification. Main results: extensive experiments are conducted on the DEAP and SEED datasets. The classification results of accuracy, precision and F-score for valence and arousal dimensions on DEAP achieved 90.67% and 90.33%, 92.38% and 91.72% and 91.34% and 90.86%, respectively. The accuracy, precision and F-score of positive, neutral and negative classification reached 94.92%, 95.34% and 94.17%, respectively on SEED dataset. Significance: the above results demonstrate the ERGL method is encouraging in comparison to the state-of-the-art recognition researches.

show abstract

Section: Experimental Results On Seedcontrasting

confidence: 51%

Spatial–temporal features-based EEG emotion recognition using graph convolution network and long short-term memory

Zheng

et al. 2023

Physiol. Meas.

View full text Add to dashboard Cite

show abstract

“…(3) Extensive experiments on SEED [27], SEED-V [28], and DEAP [2] datasets validate the effectiveness and generalizability of the proposed MSLTE model. ( 4) The proposed MSLTE model achieves higher emotion classification accuracy and training efficiency, and much lower model parameters and computational complexity than the state-of-theart (SOTA) multi-task-based model, GMSS [25].…”

Section: Introductionmentioning

confidence: 99%

“…However, most DAbased methods need to be trained on labeled training data and unlabeled test data, which does not apply in real applications. Considering that the previously mentioned EEG emotion recognition models are based on single-task learning, which may lead to overfitting and limit the generalization of the features learned by the model, Li et al [25] introduce multi-task learning by constructing three pseudo-tasks based on data augmentation, and propose a graph-based multi-task self-supervised learning (GMSS) model. However, firstly, GMSS [25] is based on data augmentation, which has a high time cost.…”

Section: Introductionmentioning

confidence: 99%

“…Considering that the previously mentioned EEG emotion recognition models are based on single-task learning, which may lead to overfitting and limit the generalization of the features learned by the model, Li et al [25] introduce multi-task learning by constructing three pseudo-tasks based on data augmentation, and propose a graph-based multi-task self-supervised learning (GMSS) model. However, firstly, GMSS [25] is based on data augmentation, which has a high time cost. Secondly, the augmentation method used by GMSS [25] is based on the disruption of the EEG channel order, so it is only applicable to specific EEG datasets, which affects its flexibility.…”

Section: Introductionmentioning

confidence: 99%

“…However, firstly, GMSS [25] is based on data augmentation, which has a high time cost. Secondly, the augmentation method used by GMSS [25] is based on the disruption of the EEG channel order, so it is only applicable to specific EEG datasets, which affects its flexibility. Thirdly, independent classifiers are adopted among multiple pseudo-tasks, which increases the model's parameters and affects the training efficiency.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

MSLTE: multiple self-supervised learning tasks for enhancing EEG emotion recognition

Li,

Chen,

Niu

et al. 2024

J. Neural Eng.

View full text Add to dashboard Cite

Objective. The instability of the EEG acquisition devices may lead to information loss in the channels or frequency bands of the collected EEG. This phenomenon may be ignored in available models, which leads to the overfitting and low generalization of the model. Approach. Multiple self-supervised learning tasks are introduced in the proposed model to enhance the generalization of EEG emotion recognition and reduce the overfitting problem to some extent. Firstly, channel masking and frequency masking are introduced to simulate the information loss in certain channels and frequency bands resulting from the instability of EEG, and two self-supervised learning-based feature reconstruction tasks combining masked graph autoencoders (GAE) are constructed to enhance the generalization of the shared encoder. Secondly, to take full advantage of the complementary information contained in these two self-supervised learning tasks to ensure the reliability of feature reconstruction, a weight sharing (WS) mechanism is introduced between the two graph decoders. Thirdly, an adaptive weight multi-task loss (AWML) strategy based on homoscedastic uncertainty is adopted to combine the supervised learning loss and the two self-supervised learning losses to enhance the performance further. Main results. Experimental results on SEED, SEED-V, and DEAP datasets demonstrate that: i) Generally, the proposed model achieves higher averaged emotion classification accuracy than various baselines included in both subject-dependent and subject-independent scenarios. ii) Each key module contributes to the performance enhancement of the proposed model. iii) It achieves higher training efficiency, and significantly lower model size and computational complexity than the state-of-the-art (SOTA) multi-task-based model. iv) The performances of the proposed model are less influenced by the key parameters. Significance. The introduction of the self-supervised learning task helps to enhance the generalization of the EEG emotion recognition model and eliminate overfitting to some extent, which can be modified to be applied in other EEG-based classification tasks.

show abstract

Clustering-based multi-featured self-supervised learning for human activities and video retrieval

Javed,

Yu,

Rajeh

et al. 2024

Appl Intell

View full text Add to dashboard Cite

GMSS: Graph-Based Multi-Task Self-Supervised Learning for EEG Emotion Recognition

Cited by 28 publications

References 49 publications

Spatial–temporal features-based EEG emotion recognition using graph convolution network and long short-term memory

Spatial–temporal features-based EEG emotion recognition using graph convolution network and long short-term memory

MSLTE: multiple self-supervised learning tasks for enhancing EEG emotion recognition

Clustering-based multi-featured self-supervised learning for human activities and video retrieval

Contact Info

Product

Resources

About