Yu Ouyang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yu Ouyang has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
- Hong Zeng
School of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou, Zhejiang, China
9 shared publications - Yaru Guo
School of Computer Science, Hangzhou Dianzi University, Hangzhou 310018, People’s Republic of China
4 shared publications - Zhongzheng Li
School of Computer Science, Hangzhou Dianzi University, Hangzhou 310018, People’s Republic of China
4 shared publications - Wenjie Cheng
School of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou, Zhejiang, China
3 shared publications - Xinni Kong
Hangzhou Dianzi University School of Computer Science and Technology, Hangzhou Dianzi University School of Computer Science and Technology, Hangzhou, Zhejiang, 310018, CHINA
3 shared publications - Lizhi Wang
School of Computer Science, Hangzhou Dianzi University, Hangzhou 310018, People’s Republic of China
2 shared publications
Latest Publications
- 2 March 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (2) , 025025
A novel framework for cognitive state identification using resting-state EEG
Recent advancements in cognitive impairment research have led to significant progress. Electroencephalography (EEG)-based cognitive state identification can detect early cognitive decline in the elderly, providing a critical window for intervention. As cognitive function worsens, neural activity patterns in the brain also change. Functional connectivity between brain regions, a key indicator of synchronized neural activity, is widely used to reveal brain network characteristics under different cognitive states. In this study, we introduce a novel framework, PowerSyncNet, based on functional connectivity to identify cognitive states. PowerSyncNet mainly consists of three modules. The Channel-Pair Feature Sequences Builder extracts features that characterize functional connectivity across different frequency bands. The Encoder4Band module captures temporal-frequency representations that reflect cognitive states and combines cross-band information to improve feature clarity. The Classifier then determines the corresponding cognitive states. We tested PowerSyncNet on the publicly available Chung-Ang University Hospital EEG (CAUEEG) dataset and our own collected Emotion and Cognition EEG (ECED) dataset. Results show that PowerSyncNet has superior cognitive identification capabilities compared with existing deep learning frameworks, facilitating early assessment and timely intervention for patients with cognitive impairment.
- 2 March 2026
- Published by IOP Publishing in Biomedical Physics & Engineering Express
- Vol. 12 (2) , 025025
A novel framework for cognitive state identification using resting-state EEG
Recent advancements in cognitive impairment research have led to significant progress. Electroencephalography (EEG)-based cognitive state identification can detect early cognitive decline in the elderly, providing a critical window for intervention. As cognitive function worsens, neural activity patterns in the brain also change. Functional connectivity between brain regions, a key indicator of synchronized neural activity, is widely used to reveal brain network characteristics under different cognitive states. In this study, we introduce a novel framework, PowerSyncNet, based on functional connectivity to identify cognitive states. PowerSyncNet mainly consists of three modules. The Channel-Pair Feature Sequences Builder extracts features that characterize functional connectivity across different frequency bands. The Encoder4Band module captures temporal-frequency representations that reflect cognitive states and combines cross-band information to improve feature clarity. The Classifier then determines the corresponding cognitive states. We tested PowerSyncNet on the publicly available Chung-Ang University Hospital EEG (CAUEEG) dataset and our own collected Emotion and Cognition EEG (ECED) dataset. Results show that PowerSyncNet has superior cognitive identification capabilities compared with existing deep learning frameworks, facilitating early assessment and timely intervention for patients with cognitive impairment.
- 22 January 2026
- Published by IOP Publishing in Journal of Neural Engineering
- Vol. 23 (1) , 016013
MMoGCN: a multi-gate mixture of graph convolutional network model for EEG emotion and mood disorder recognition
Objective. Emotional states and mood disorders are closely interconnected, and their joint recognition serves as a critical pathway to uncovering their intrinsic relationship. Currently, deep learning (DL) models based on electroencephalogram (EEG) have achieved significant progress in single tasks such as emotion recognition or mood disorder (MD) recognition. However, most existing models are limited to handling only one of these tasks independently and fail to effectively leverage the shared features in EEG data related to both emotions and mood disorders. This limitation hinders the in-depth exploration of the complex interplay between emotions and mood disorders. Therefore, this study aims to develop an EEG-based DL framework for the joint recognition of emotions and mood disorders, thereby providing a foundation for further investigation into their interaction. Approach. We design a multi-gate mixture-of-experts graph convolutional network model(MMoGCN) for joint emotion and MD recognition. MMoGCN comprises three key modules: (1) a feature extraction module based on differential entropy to robustly represent EEG signals; (2) a Multi-gated shared experts module, which integrates two experts, and combines them through a gating mechanism to extract shared representations across tasks; and (3) adaptive task-specific towers, which consist of individual classification towers for each task and incorporate an adaptive weighting loss function to dynamically adjust task contributions. MMoGCN is evaluated on a self-collected dataset and further validated on the public DEAP dataset. Main results. MMoGCN achieves superior performance compared with state-of-the-art single-task and multi-task baselines in both emotion and MD recognition. Validation experiments on DEAP further demonstrate the scalability and generalization of MMoGCN. Significance. An effective multi-task learning model is proposed for joint emotion and MD recognition based on EEG. Additionally, the cognitive differences are also analyzed in emotional responses between healthy controls and subjects with mood disorders, providing methodological insights and potential assistance for cognitive rehabilitation from both cognitive and emotional perspectives.
- 22 January 2026
- Published by IOP Publishing in Journal of Neural Engineering
- Vol. 23 (1) , 016013
MMoGCN: a multi-gate mixture of graph convolutional network model for EEG emotion and mood disorder recognition
Objective. Emotional states and mood disorders are closely interconnected, and their joint recognition serves as a critical pathway to uncovering their intrinsic relationship. Currently, deep learning (DL) models based on electroencephalogram (EEG) have achieved significant progress in single tasks such as emotion recognition or mood disorder (MD) recognition. However, most existing models are limited to handling only one of these tasks independently and fail to effectively leverage the shared features in EEG data related to both emotions and mood disorders. This limitation hinders the in-depth exploration of the complex interplay between emotions and mood disorders. Therefore, this study aims to develop an EEG-based DL framework for the joint recognition of emotions and mood disorders, thereby providing a foundation for further investigation into their interaction. Approach. We design a multi-gate mixture-of-experts graph convolutional network model(MMoGCN) for joint emotion and MD recognition. MMoGCN comprises three key modules: (1) a feature extraction module based on differential entropy to robustly represent EEG signals; (2) a Multi-gated shared experts module, which integrates two experts, and combines them through a gating mechanism to extract shared representations across tasks; and (3) adaptive task-specific towers, which consist of individual classification towers for each task and incorporate an adaptive weighting loss function to dynamically adjust task contributions. MMoGCN is evaluated on a self-collected dataset and further validated on the public DEAP dataset. Main results. MMoGCN achieves superior performance compared with state-of-the-art single-task and multi-task baselines in both emotion and MD recognition. Validation experiments on DEAP further demonstrate the scalability and generalization of MMoGCN. Significance. An effective multi-task learning model is proposed for joint emotion and MD recognition based on EEG. Additionally, the cognitive differences are also analyzed in emotional responses between healthy controls and subjects with mood disorders, providing methodological insights and potential assistance for cognitive rehabilitation from both cognitive and emotional perspectives.
- Open Access
- 5 November 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (21) , 11777
Machine Learning, Physiological Signals, and Emotional Stress/Anxiety: Pitfalls and Challenges
Anxiety and emotional stress are pervasive psychological challenges that profoundly impact human health in today’s fast-paced society. Traditional assessment methods, such as self-reports and clinical interviews, often suffer from subjective biases and lack the capability for objective, real-time evaluation of mental states. However, the integration of physiological signals—including electroencephalography (EEG), heart rate (HR), electrodermal activity (EDA), and eye movements—with advanced machine learning (ML) techniques, offers a promising approach to automate and objectify mental health assessments. A systematic review was conducted to explore recent advances in the early detection of anxiety and stress by combining physiological signals and ML methods. To assess methodological quality, a specific analysis framework was designed for the 113 studies included, which identified significant deficiencies in the literature. This highlights the urgent need to adopt standardized reporting guidelines in this field. The role of these technologies in feature extraction, classification, and predictive modeling was analyzed, also addressing critical challenges related to data quality, model interpretability, and the influence of intersectional factors like gender and age. Ethical and privacy considerations in the current research were also included. Finally, potential avenues for future research were summarized, highlighting the potential of ML technologies for early detection and proactive intervention in mental disorders.
- 5 November 2025
- Published by MDPI AG in Applied Sciences
- Vol. 15 (21) , 11777
Open AccessMachine Learning, Physiological Signals, and Emotional Stress/Anxiety: Pitfalls and Challenges
Anxiety and emotional stress are pervasive psychological challenges that profoundly impact human health in today’s fast-paced society. Traditional assessment methods, such as self-reports and clinical interviews, often suffer from subjective biases and lack the capability for objective, real-time evaluation of mental states. However, the integration of physiological signals—including electroencephalography (EEG), heart rate (HR), electrodermal activity (EDA), and eye movements—with advanced machine learning (ML) techniques, offers a promising approach to automate and objectify mental health assessments. A systematic review was conducted to explore recent advances in the early detection of anxiety and stress by combining physiological signals and ML methods. To assess methodological quality, a specific analysis framework was designed for the 113 studies included, which identified significant deficiencies in the literature. This highlights the urgent need to adopt standardized reporting guidelines in this field. The role of these technologies in feature extraction, classification, and predictive modeling was analyzed, also addressing critical challenges related to data quality, model interpretability, and the influence of intersectional factors like gender and age. Ethical and privacy considerations in the current research were also included. Finally, potential avenues for future research were summarized, highlighting the potential of ML technologies for early detection and proactive intervention in mental disorders.
- 9 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 30 (4) , 3529-3542
P3DL: A Privacy Preserving Personalized Distributed Learning Framework for EEG-Based Cognitive State Identification
Electroencephalography (EEG)-based brain cognitive state identification for the elderly allows timely detection and early intervention of cognitive deterioration. Notably, EEG signals carry a great deal of vital personal information. However, a majority of the existing cognitive evaluations focus on improving the accuracy of EEG decoding and enhancing the performance of identification models, while neglecting the privacy protection of EEG data. To address the risky challenge, we propose a privacy-preserving personalized distributed learning framework (P3DL) for cognitive state identification. Specifically, it consists of the clients and a central server. Each client contains a cognitive model and a score model for identifying cognitive states and quantifying cognitive levels, respectively. The central server can aggregate local models' parameters from distributed clients, then, update and downstream the global model's parameters for iterative optimization. A federated dynamic update strategy (FedDBS) is designed to jointly update all global and local models with a supervisory metric. In order to further improve the identification performance and judge the misdiagnosis level, a novel loss function, extreme error Loss (E2Loss), is proposed. Compared with the baseline, experimental results on our self-collected clinical dataset and a public dataset show an average increase in F2Score of 5.58% and 3.31%, and in accuracy of 1.78% and 2.46%, respectively. Furthermore, the scalability of the framework has been proved in the emotion recognition task. Our proposed framework P3DL can not only improve the identification performance, but also protect the privacy of EEG, opening a new window for secure healthcare.
- 9 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 30 (4) , 3529-3542
P3DL: A Privacy Preserving Personalized Distributed Learning Framework for EEG-Based Cognitive State Identification
Electroencephalography (EEG)-based brain cognitive state identification for the elderly allows timely detection and early intervention of cognitive deterioration. Notably, EEG signals carry a great deal of vital personal information. However, a majority of the existing cognitive evaluations focus on improving the accuracy of EEG decoding and enhancing the performance of identification models, while neglecting the privacy protection of EEG data. To address the risky challenge, we propose a privacy-preserving personalized distributed learning framework (P3DL) for cognitive state identification. Specifically, it consists of the clients and a central server. Each client contains a cognitive model and a score model for identifying cognitive states and quantifying cognitive levels, respectively. The central server can aggregate local models' parameters from distributed clients, then, update and downstream the global model's parameters for iterative optimization. A federated dynamic update strategy (FedDBS) is designed to jointly update all global and local models with a supervisory metric. In order to further improve the identification performance and judge the misdiagnosis level, a novel loss function, extreme error Loss (E2Loss), is proposed. Compared with the baseline, experimental results on our self-collected clinical dataset and a public dataset show an average increase in F2Score of 5.58% and 3.31%, and in accuracy of 1.78% and 2.46%, respectively. Furthermore, the scalability of the framework has been proved in the emotion recognition task. Our proposed framework P3DL can not only improve the identification performance, but also protect the privacy of EEG, opening a new window for secure healthcare.
- 17 May 2025
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 268, 108835
MT-RCAF: A Multi-Task Residual Cross Attention Framework for EEG-based emotion recognition and mood disorder detection
- 17 May 2025
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 268, 108835
MT-RCAF: A Multi-Task Residual Cross Attention Framework for EEG-based emotion recognition and mood disorder detection
- 25 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
A Novel Dual-Task Model for EEG-Based Emotion and Cognition Recognition
Human cognition and emotion states are interrelated. Constructing EEG-based deep learning models to simultaneously recognize the cognitive and emotional states of an individual contributes to cognitive and emotional interaction studies. However, most of the existing models focus on cognition or emotion recognition, neglecting to exploit their shared features in EEG and weakening the generalization performance of the models. In this study, we propose a dual-task learning model DT-EEGNet, where emotion and cognition are considered as related tasks for joint analysis. In DT-EEGNet, EEGNet is employed as a base network responsible for extracting shared features between cognition and emotion recognition tasks. Then a multi-scale ECAWeight attention (MSEA) module is introduced for obtaining key information from these shared features. Finally, a dual-task loss function named Dynamic Weight Average (DWA) is used to balance the training rates of emotion and cognition recognition tasks for better overall training performance. Experimental results on our self-constructed emotion and cognition EEG dataset (ECED) and the public DEAP dataset show that our proposed DT-EEGNet has better dual-task recognition performance. The proposed method can provide a new idea for cognitive impairment assessment research as well.
- 25 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
A Novel Dual-Task Model for EEG-Based Emotion and Cognition Recognition
Human cognition and emotion states are interrelated. Constructing EEG-based deep learning models to simultaneously recognize the cognitive and emotional states of an individual contributes to cognitive and emotional interaction studies. However, most of the existing models focus on cognition or emotion recognition, neglecting to exploit their shared features in EEG and weakening the generalization performance of the models. In this study, we propose a dual-task learning model DT-EEGNet, where emotion and cognition are considered as related tasks for joint analysis. In DT-EEGNet, EEGNet is employed as a base network responsible for extracting shared features between cognition and emotion recognition tasks. Then a multi-scale ECAWeight attention (MSEA) module is introduced for obtaining key information from these shared features. Finally, a dual-task loss function named Dynamic Weight Average (DWA) is used to balance the training rates of emotion and cognition recognition tasks for better overall training performance. Experimental results on our self-constructed emotion and cognition EEG dataset (ECED) and the public DEAP dataset show that our proposed DT-EEGNet has better dual-task recognition performance. The proposed method can provide a new idea for cognitive impairment assessment research as well.
- 7 October 2024
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 100
DAEEGViT: A domain adaptive vision transformer framework for EEG cognitive state identification
- 7 October 2024
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 100
DAEEGViT: A domain adaptive vision transformer framework for EEG cognitive state identification
- 23 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-14
ARFN: An Attention-Based Recurrent Fuzzy Network for EEG Mental Workload Assessment
Assessing mental workload using electroencephalogram (EEG) signals is a significant research avenue within the brain-computer interface domain. However, due to the low signal-to-noise ratio in EEG signals and the inter-individual variability in EEG data acquisition, achieving high accuracy and generalization in feature extraction and classification for mental workload assessment is still challenging. We propose a novel deep learning framework named attention-based recurrent fuzzy network (ARFN) for EEG mental workload assessment. In ARFN, we adopt a fuzzy recursive module which employs a feature attention mechanism and a fuzzy rule attention mechanism, respectively, to flexibly extract EEG features related to mental workload. The former can extract the frequency domain features of EEG signals, while the latter is used to represent the membership degrees within the distribution of frequency features, so as to find effective fuzzy rules for classification. Subsequently, the output of the fuzzy recursive module is directed into the long short-term memory (LSTM) to further extract temporal features of EEG, followed by a fully connected layer and the Softmax function for classification. The experimental results on three public datasets show that ARFN outperforms other state-of-the-art models in EEG mental workload assessment.
- 23 February 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-14
ARFN: An Attention-Based Recurrent Fuzzy Network for EEG Mental Workload Assessment
Assessing mental workload using electroencephalogram (EEG) signals is a significant research avenue within the brain-computer interface domain. However, due to the low signal-to-noise ratio in EEG signals and the inter-individual variability in EEG data acquisition, achieving high accuracy and generalization in feature extraction and classification for mental workload assessment is still challenging. We propose a novel deep learning framework named attention-based recurrent fuzzy network (ARFN) for EEG mental workload assessment. In ARFN, we adopt a fuzzy recursive module which employs a feature attention mechanism and a fuzzy rule attention mechanism, respectively, to flexibly extract EEG features related to mental workload. The former can extract the frequency domain features of EEG signals, while the latter is used to represent the membership degrees within the distribution of frequency features, so as to find effective fuzzy rules for classification. Subsequently, the output of the fuzzy recursive module is directed into the long short-term memory (LSTM) to further extract temporal features of EEG, followed by a fully connected layer and the Softmax function for classification. The experimental results on three public datasets show that ARFN outperforms other state-of-the-art models in EEG mental workload assessment.
- Top Cited Papers
- 20 November 2023
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 243, 107927
EESCN: A novel spiking neural network method for EEG-based emotion recognition
- 20 November 2023
- Published by Elsevier in Computer Methods and Programs in Biomedicine
- Vol. 243, 107927
Top Cited PapersEESCN: A novel spiking neural network method for EEG-based emotion recognition