Gang Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Gang Chen has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
Shengya Shen
School of Information Engineering, Zhengzhou University, Zhengzhou 450000, China
5 shared publications- Zongmin Wang
School of Water Conservancy and Transportation, Zhengzhou University, Zhengzhou 450001, China
5 shared publications - Runchuan Li
School of Information Engineering, Zhengzhou University, Zhengzhou 450000, China
4 shared publications - Bing Zhou
The College of Computer Science and Artificial Intelligence, Zhengzhou University, Zhengzhou, China
3 shared publications - Deliang Jin
School of Computer Science and Artificial Intelligence, Zhengzhou University, Zhengzhou 450001, China
3 shared publications - Runzhi Li
Zhengzhou University
3 shared publications
Latest Publications
- Open Access
- 5 September 2025
- Published by MDPI AG in Machine Learning and Knowledge Extraction
- Vol. 7 (3) , 95
Machine Unlearning for Robust DNNs: Attribution-Guided Partitioning and Neuron Pruning in Noisy Environments
Deep neural networks (DNNs) are highly effective across many domains but are sensitive to noisy or corrupted training data. Existing noise mitigation strategies often rely on strong assumptions about noise distributions or require costly retraining, limiting their scalability. Inspired by machine unlearning, we propose a novel framework that integrates attribution-guided data partitioning, neuron pruning, and targeted fine-tuning to enhance robustness. Our method uses gradient-based attribution to probabilistically identify clean samples without assuming specific noise characteristics. It then applies sensitivity-based neuron pruning to remove components most susceptible to noise, followed by fine-tuning on the retained high-quality subset. This approach jointly addresses data and model-level noise, offering a practical alternative to full retraining or explicit noise modeling. We evaluate our method on CIFAR-10 image classification and keyword spotting tasks under varying levels of label corruption. On CIFAR-10, our framework improves accuracy by up to 10% (F-FT vs. retrain) and reduces retraining time by 47% (L-FT vs. retrain), highlighting both accuracy and efficiency gains. These results highlight its effectiveness and efficiency in noisy settings, making it a scalable solution for robust generalization.
- 5 September 2025
- Published by MDPI AG in Machine Learning and Knowledge Extraction
- Vol. 7 (3) , 95
Open AccessMachine Unlearning for Robust DNNs: Attribution-Guided Partitioning and Neuron Pruning in Noisy Environments
Deep neural networks (DNNs) are highly effective across many domains but are sensitive to noisy or corrupted training data. Existing noise mitigation strategies often rely on strong assumptions about noise distributions or require costly retraining, limiting their scalability. Inspired by machine unlearning, we propose a novel framework that integrates attribution-guided data partitioning, neuron pruning, and targeted fine-tuning to enhance robustness. Our method uses gradient-based attribution to probabilistically identify clean samples without assuming specific noise characteristics. It then applies sensitivity-based neuron pruning to remove components most susceptible to noise, followed by fine-tuning on the retained high-quality subset. This approach jointly addresses data and model-level noise, offering a practical alternative to full retraining or explicit noise modeling. We evaluate our method on CIFAR-10 image classification and keyword spotting tasks under varying levels of label corruption. On CIFAR-10, our framework improves accuracy by up to 10% (F-FT vs. retrain) and reduces retraining time by 47% (L-FT vs. retrain), highlighting both accuracy and efficiency gains. These results highlight its effectiveness and efficiency in noisy settings, making it a scalable solution for robust generalization.
- Open Access
Machine Unlearning for Robust DNNs: Attribution-Guided Partitioning and Neuron Pruning in Noisy Environments
Deep neural networks (DNNs) have achieved remarkable success across diverse domains, but their performance can be severely degraded by noisy or corrupted training data. Conventional noise mitigation methods often rely on explicit assumptions about noise distributions or require extensive retraining, which can be impractical for large-scale models. Inspired by the principles of machine unlearning, we propose a novel framework that integrates attribution-guided data partitioning, discriminative neuron pruning, and targeted fine-tuning to mitigate the impact of noisy samples. Our approach employs gradient-based attribution to probabilistically distinguish high-quality examples from potentially corrupted ones without imposing restrictive assumptions on the noise. It then applies regression-based sensitivity analysis to identify and prune neurons that are most vulnerable to noise. Finally, the resulting network is fine-tuned on the high-quality data subset to efficiently recover and enhance its generalization performance. This integrated unlearning-inspired framework provides several advantages over conventional noise-robust learning approaches. Notably, it combines data-level unlearning with model-level adaptation, thereby avoiding the need for full model retraining or explicit noise modeling. We evaluate our method on representative tasks (e.g., CIFAR-10 image classification and speech recognition) under various noise levels and observe substantial gains in both accuracy and efficiency. For example, our framework achieves approximately a 10% absolute accuracy improvement over standard retraining on CIFAR-10 with injected label noise, while reducing retraining time by up to 47% in some settings. These results demonstrate the effectiveness and scalability of the proposed approach for achieving robust generalization in noisy environments.
Open AccessMachine Unlearning for Robust DNNs: Attribution-Guided Partitioning and Neuron Pruning in Noisy Environments
Deep neural networks (DNNs) have achieved remarkable success across diverse domains, but their performance can be severely degraded by noisy or corrupted training data. Conventional noise mitigation methods often rely on explicit assumptions about noise distributions or require extensive retraining, which can be impractical for large-scale models. Inspired by the principles of machine unlearning, we propose a novel framework that integrates attribution-guided data partitioning, discriminative neuron pruning, and targeted fine-tuning to mitigate the impact of noisy samples. Our approach employs gradient-based attribution to probabilistically distinguish high-quality examples from potentially corrupted ones without imposing restrictive assumptions on the noise. It then applies regression-based sensitivity analysis to identify and prune neurons that are most vulnerable to noise. Finally, the resulting network is fine-tuned on the high-quality data subset to efficiently recover and enhance its generalization performance. This integrated unlearning-inspired framework provides several advantages over conventional noise-robust learning approaches. Notably, it combines data-level unlearning with model-level adaptation, thereby avoiding the need for full model retraining or explicit noise modeling. We evaluate our method on representative tasks (e.g., CIFAR-10 image classification and speech recognition) under various noise levels and observe substantial gains in both accuracy and efficiency. For example, our framework achieves approximately a 10% absolute accuracy improvement over standard retraining on CIFAR-10 with injected label noise, while reducing retraining time by up to 47% in some settings. These results demonstrate the effectiveness and scalability of the proposed approach for achieving robust generalization in noisy environments.
- 7 May 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal of Pattern Recognition and Artificial Intelligence
- Vol. 39 (06)
An Enhanced Website Fingerprinting Attacks Based on Attention Mechanism and Spatial-Temporal Features
Website fingerprinting (WF) attacks leverage deep learning to analyze encrypted traffic and identify the websites accessed by Tor users, posing a critical threat to online anonymity. However, existing detection models often suffer from low accuracy and weak robustness in open-world scenarios. To address these issues, we propose a novel spatiotemporal WF framework integrating a lightweight attention mechanism and parallel multi-size feature extraction with dilated convolution. This design significantly enhances detection performance while maintaining high efficiency. Experimental results show that our approach outperforms CNN and LSTM models by 3–5% in accuracy and surpasses the DF model by about 1%, while achieving a 50% increase in efficiency. Furthermore, it demonstrates robust performance as the number of website categories grows and remains effective against defenses such as WTF-PAD and Walkie-Talkie. Notably, our framework also mitigates the impact of concept drift, exhibiting minimal performance degradation when traffic distributions evolve over time. These findings not only underscore the unique potential of our model to advance WF attacks, but also highlight the urgent need for stronger privacy-preserving measures to protect Tor users in real-world environments.
- 7 May 2025
- Published by World Scientific Pub Co Pte Ltd in International Journal of Pattern Recognition and Artificial Intelligence
- Vol. 39 (06)
An Enhanced Website Fingerprinting Attacks Based on Attention Mechanism and Spatial-Temporal Features
Website fingerprinting (WF) attacks leverage deep learning to analyze encrypted traffic and identify the websites accessed by Tor users, posing a critical threat to online anonymity. However, existing detection models often suffer from low accuracy and weak robustness in open-world scenarios. To address these issues, we propose a novel spatiotemporal WF framework integrating a lightweight attention mechanism and parallel multi-size feature extraction with dilated convolution. This design significantly enhances detection performance while maintaining high efficiency. Experimental results show that our approach outperforms CNN and LSTM models by 3–5% in accuracy and surpasses the DF model by about 1%, while achieving a 50% increase in efficiency. Furthermore, it demonstrates robust performance as the number of website categories grows and remains effective against defenses such as WTF-PAD and Walkie-Talkie. Notably, our framework also mitigates the impact of concept drift, exhibiting minimal performance degradation when traffic distributions evolve over time. These findings not only underscore the unique potential of our model to advance WF attacks, but also highlight the urgent need for stronger privacy-preserving measures to protect Tor users in real-world environments.
- Open Access
- 22 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 1
Static and multivariate-temporal attentive fusion transformer for readmission risk prediction of ICU patients
Accurate short-term readmission prediction of ICU patients is significant in improving the efficiency of resource assignment. Clinically, both individual static and multivariate-temporal data play critical roles in short-term readmission prediction. It proposes a novel static and multivariate-temporal attentive fusion transformer (SMTAFormer) to predict short-term readmission of ICU patients by fully leveraging the potential of demographic and dynamic temporal data. In SMTAFormer, it first applies an MLP network and a temporal transformer network to learn useful static and temporal feature representations, respectively. Then, the well designed static and multivariate-temporal feature fusion module is applied to fuse static and temporal feature representations by modeling intra-correlation among multivariate-temporal features and constructing inter-correlation between static and multivariate temporal features. It constructs a readmission risk assessment of ICU patients (RRA) dataset based on the MIMIC-III dataset. The extensive experiments show that SMTAFormer outperforms advanced methods, in which the accuracy of the proposed method is up to 86.6 %. The proposed SMTAFormer can efficiently capture and fuse static and multivariate temporal feature representations.
- 22 November 2024
- Published by Association for Computing Machinery (ACM)
- p. 1
Open AccessStatic and multivariate-temporal attentive fusion transformer for readmission risk prediction of ICU patients
Accurate short-term readmission prediction of ICU patients is significant in improving the efficiency of resource assignment. Clinically, both individual static and multivariate-temporal data play critical roles in short-term readmission prediction. It proposes a novel static and multivariate-temporal attentive fusion transformer (SMTAFormer) to predict short-term readmission of ICU patients by fully leveraging the potential of demographic and dynamic temporal data. In SMTAFormer, it first applies an MLP network and a temporal transformer network to learn useful static and temporal feature representations, respectively. Then, the well designed static and multivariate-temporal feature fusion module is applied to fuse static and temporal feature representations by modeling intra-correlation among multivariate-temporal features and constructing inter-correlation between static and multivariate temporal features. It constructs a readmission risk assessment of ICU patients (RRA) dataset based on the MIMIC-III dataset. The extensive experiments show that SMTAFormer outperforms advanced methods, in which the accuracy of the proposed method is up to 86.6 %. The proposed SMTAFormer can efficiently capture and fuse static and multivariate temporal feature representations.
- Open Access
Static and multivariate-temporal attentive fusion transformer for readmission risk prediction
Background: Accurate short-term readmission prediction of ICU patients is significant in improving the efficiency of resource assignment by assisting physicians in making discharge decisions. Clinically, both individual static static and multivariate temporal data collected from ICU monitors play critical roles in short-term readmission prediction. Informative static and multivariate temporal feature representation capturing and fusion present challenges for accurate readmission prediction. Methods:We propose a novel static and multivariate-temporal attentive fusion transformer (SMTAFormer) to predict short-term readmission of ICU patients by fully leveraging the potential of demographic and dynamic temporal data. In SMTAFormer, we first apply an MLP network and a temporal transformer network to learn useful static and temporal feature representations, respectively. Then, the well-designed static and multivariate temporal feature fusion module is applied to fuse static and temporal feature representations by modeling intra-correlation among multivariate temporal features and constructing inter-correlation between static and multivariate temporal features. Results: We construct a readmission risk assessment (RRA) dataset based on the MIMIC-III dataset. The extensive experiments show that SMTAFormer outperforms advanced methods, in which the accuracy of our proposed method is up to 86.6%, and the area under the receiver operating characteristic curve (AUC) is up to 0.717. Conclusion: Our proposed SMTAFormer can efficiently capture and fuse static and multivariate temporal feature representations. The results show that SMTAFormer significantly improves the short-term readmission prediction performance of ICU patients through comparisons to strong baselines.
Open AccessStatic and multivariate-temporal attentive fusion transformer for readmission risk prediction
Background: Accurate short-term readmission prediction of ICU patients is significant in improving the efficiency of resource assignment by assisting physicians in making discharge decisions. Clinically, both individual static static and multivariate temporal data collected from ICU monitors play critical roles in short-term readmission prediction. Informative static and multivariate temporal feature representation capturing and fusion present challenges for accurate readmission prediction. Methods:We propose a novel static and multivariate-temporal attentive fusion transformer (SMTAFormer) to predict short-term readmission of ICU patients by fully leveraging the potential of demographic and dynamic temporal data. In SMTAFormer, we first apply an MLP network and a temporal transformer network to learn useful static and temporal feature representations, respectively. Then, the well-designed static and multivariate temporal feature fusion module is applied to fuse static and temporal feature representations by modeling intra-correlation among multivariate temporal features and constructing inter-correlation between static and multivariate temporal features. Results: We construct a readmission risk assessment (RRA) dataset based on the MIMIC-III dataset. The extensive experiments show that SMTAFormer outperforms advanced methods, in which the accuracy of our proposed method is up to 86.6%, and the area under the receiver operating characteristic curve (AUC) is up to 0.717. Conclusion: Our proposed SMTAFormer can efficiently capture and fuse static and multivariate temporal feature representations. The results show that SMTAFormer significantly improves the short-term readmission prediction performance of ICU patients through comparisons to strong baselines.
- Open Access
- 22 January 2022
- Published by MDPI AG in International Journal of Environmental Research and Public Health
- Vol. 19 (3) , 1237
Predicting Continuity of Asthma Care Using a Machine Learning Model: Retrospective Cohort Study
Continuity of care (COC) has been shown to possess numerous health benefits for chronic diseases. Specifically, the establishment of its level can facilitate clinical decision-making and enhanced allocation of healthcare resources. However, the use of a generalizable predictive methodology to determine the COC in patients has been underinvestigated. To fill this research gap, this study aimed to develop a machine learning model to predict the future COC of asthma patients and explore the associated factors. We included 31,724 adult outpatients with asthma who received care from the University of Washington Medicine between 2011 and 2018, and examined 138 features to build the machine learning model. Following the 10-fold cross-validations, the proposed model yielded an accuracy of 88.20%, an average area under the receiver operating characteristic curve of 0.96, and an average F1 score of 0.86. Further analysis revealed that the severity of asthma, comorbidities, insurance, and age were highly correlated with the COC of patients with asthma. This study used predictive methods to obtain the COC of patients, and our excellent modeling strategy achieved high performance. After further optimization, the model could facilitate future clinical decisions, hospital management, and improve outcomes.
- 22 January 2022
- Published by MDPI AG in International Journal of Environmental Research and Public Health
- Vol. 19 (3) , 1237
Open AccessPredicting Continuity of Asthma Care Using a Machine Learning Model: Retrospective Cohort Study
Continuity of care (COC) has been shown to possess numerous health benefits for chronic diseases. Specifically, the establishment of its level can facilitate clinical decision-making and enhanced allocation of healthcare resources. However, the use of a generalizable predictive methodology to determine the COC in patients has been underinvestigated. To fill this research gap, this study aimed to develop a machine learning model to predict the future COC of asthma patients and explore the associated factors. We included 31,724 adult outpatients with asthma who received care from the University of Washington Medicine between 2011 and 2018, and examined 138 features to build the machine learning model. Following the 10-fold cross-validations, the proposed model yielded an accuracy of 88.20%, an average area under the receiver operating characteristic curve of 0.96, and an average F1 score of 0.86. Further analysis revealed that the severity of asthma, comorbidities, insurance, and age were highly correlated with the COC of patients with asthma. This study used predictive methods to obtain the COC of patients, and our excellent modeling strategy achieved high performance. After further optimization, the model could facilitate future clinical decisions, hospital management, and improve outcomes.
- Open Access
- 13 December 2021
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2021, 1-16
Combining Rhythm Information between Heartbeats and BiLSTM-Treg Algorithm for Intelligent Beat Classification of Arrhythmia
Arrhythmia is a cardiovascular disease that seriously affects human health. The identification and diagnosis of arrhythmia is an effective means of preventing most heart diseases. In this paper, a BiLSTM-Treg algorithm that integrates rhythm information is proposed to realize the automatic classification of arrhythmia. Firstly, the discrete wavelet transform is used to denoise the ECG signal, based on which we performed heartbeat segmentation and preserved the timing relationship between heartbeats. Then, different heartbeat segment lengths and the BiLSTM network model are used to conduct multiple experiments to select the optimal heartbeat segment length. Finally, the tree regularization method is used to optimize the BiLSTM network model to improve classification accuracy. And the interpretability of the neural network model is analyzed by analyzing the simulated decision tree generated in the tree regularization method. This method divides the heartbeat into five categories (nonectopic (N), supraventricular ectopic (S), ventricular ectopic (V), fused heartbeats (F), and unknown heartbeats (Q)) and is validated on the MIT-BIH arrhythmia database. The results show that the overall classification accuracy of the algorithm is 99.32%. Compared with other methods of classifying heartbeat, the BiLSTM-Treg network model algorithm proposed in this paper not only improves the classification accuracy and obtains higher sensitivity and positive predictive value but also has higher interpretability.
- 13 December 2021
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2021, 1-16
Open AccessCombining Rhythm Information between Heartbeats and BiLSTM-Treg Algorithm for Intelligent Beat Classification of Arrhythmia
Arrhythmia is a cardiovascular disease that seriously affects human health. The identification and diagnosis of arrhythmia is an effective means of preventing most heart diseases. In this paper, a BiLSTM-Treg algorithm that integrates rhythm information is proposed to realize the automatic classification of arrhythmia. Firstly, the discrete wavelet transform is used to denoise the ECG signal, based on which we performed heartbeat segmentation and preserved the timing relationship between heartbeats. Then, different heartbeat segment lengths and the BiLSTM network model are used to conduct multiple experiments to select the optimal heartbeat segment length. Finally, the tree regularization method is used to optimize the BiLSTM network model to improve classification accuracy. And the interpretability of the neural network model is analyzed by analyzing the simulated decision tree generated in the tree regularization method. This method divides the heartbeat into five categories (nonectopic (N), supraventricular ectopic (S), ventricular ectopic (V), fused heartbeats (F), and unknown heartbeats (Q)) and is validated on the MIT-BIH arrhythmia database. The results show that the overall classification accuracy of the algorithm is 99.32%. Compared with other methods of classifying heartbeat, the BiLSTM-Treg network model algorithm proposed in this paper not only improves the classification accuracy and obtains higher sensitivity and positive predictive value but also has higher interpretability.
- Open Access
- 12 October 2021
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2021, 1-15
Interpretable Detection and Location of Myocardial Infarction Based on Ventricular Fusion Rule Features
Myocardial infarction (MI) is one of the most common cardiovascular diseases threatening human life. In order to accurately distinguish myocardial infarction and have a good interpretability, the classification method that combines rule features and ventricular activity features is proposed in this paper. Specifically, according to the clinical diagnosis rule and the pathological changes of myocardial infarction on the electrocardiogram, the local information extracted from the Q wave, ST segment, and T wave is computed as the rule feature. All samples of the QT segment are extracted as ventricular activity features. Then, in order to reduce the computational complexity of the ventricular activity features, the effects of Discrete Wavelet Transform (DWT), Principal Component Analysis (PCA), and Locality Preserving Projections (LPP) on the extracted ventricular activity features are compared. Combining rule features and ventricular activity features, all the 12 leads features are fused as the ultimate feature vector. Finally, eXtreme Gradient Boosting (XGBoost) is used to identify myocardial infarction, and the overall accuracy rate of 99.86% is obtained on the Physikalisch-Technische Bundesanstalt (PTB) database. This method has a good medical diagnosis basis while improving the accuracy, which is very important for clinical decision-making.
- 12 October 2021
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2021, 1-15
Open AccessInterpretable Detection and Location of Myocardial Infarction Based on Ventricular Fusion Rule Features
Myocardial infarction (MI) is one of the most common cardiovascular diseases threatening human life. In order to accurately distinguish myocardial infarction and have a good interpretability, the classification method that combines rule features and ventricular activity features is proposed in this paper. Specifically, according to the clinical diagnosis rule and the pathological changes of myocardial infarction on the electrocardiogram, the local information extracted from the Q wave, ST segment, and T wave is computed as the rule feature. All samples of the QT segment are extracted as ventricular activity features. Then, in order to reduce the computational complexity of the ventricular activity features, the effects of Discrete Wavelet Transform (DWT), Principal Component Analysis (PCA), and Locality Preserving Projections (LPP) on the extracted ventricular activity features are compared. Combining rule features and ventricular activity features, all the 12 leads features are fused as the ultimate feature vector. Finally, eXtreme Gradient Boosting (XGBoost) is used to identify myocardial infarction, and the overall accuracy rate of 99.86% is obtained on the Physikalisch-Technische Bundesanstalt (PTB) database. This method has a good medical diagnosis basis while improving the accuracy, which is very important for clinical decision-making.
- Open Access
- 9 July 2021
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2021, 1-19
An Intelligent Heartbeat Classification System Based on Attributable Features with AdaBoost+Random Forest Algorithm
Arrhythmia is a common cardiovascular disease that can threaten human life. In order to assist doctors in accurately diagnosing arrhythmia, an intelligent heartbeat classification system based on the selected optimal feature sets and AdaBoost + Random Forest model is developed. This system can acquire ECG signals through the Holter and transmit them to the cloud platform for preprocessing and feature extraction, and the features are input into AdaBoost + Random Forest for heartbeat classification. The analysis results are output in the form of reports. In this system, by comparing and analyzing the classification accuracy of different feature sets and classifiers, the optimal classification algorithm is obtained and applied to the system. The algorithm accuracy of the system is tested based on the MIT-BIH data set. The result shows that AdaBoost + Random Forest achieved 99.11% accuracy with optimal feature sets. The intelligent heartbeat classification system based on this algorithm has also achieved good results on clinical data.
- 9 July 2021
- Published by Wiley in Journal of Healthcare Engineering
- Vol. 2021, 1-19
Open AccessAn Intelligent Heartbeat Classification System Based on Attributable Features with AdaBoost+Random Forest Algorithm
Arrhythmia is a common cardiovascular disease that can threaten human life. In order to assist doctors in accurately diagnosing arrhythmia, an intelligent heartbeat classification system based on the selected optimal feature sets and AdaBoost + Random Forest model is developed. This system can acquire ECG signals through the Holter and transmit them to the cloud platform for preprocessing and feature extraction, and the features are input into AdaBoost + Random Forest for heartbeat classification. The analysis results are output in the form of reports. In this system, by comparing and analyzing the classification accuracy of different feature sets and classifiers, the optimal classification algorithm is obtained and applied to the system. The algorithm accuracy of the system is tested based on the MIT-BIH data set. The result shows that AdaBoost + Random Forest achieved 99.11% accuracy with optimal feature sets. The intelligent heartbeat classification system based on this algorithm has also achieved good results on clinical data.
- Open Access
- 25 May 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 78174-78187
Testing the Feasibility of a Multi-Model Fusion Method for Monitoring the Action of Rehabilitating Stroke Patients in Care Management
Post-stroke care encounters challenges, including high cost, lack of professionals, and insufficient rehabilitation state evaluation. Computer technology can alleviate these issues, as it allows health care professionals (HCP) to quantify the workload and thus enhance rehabilitation care quality. In this paper, a novel multi-model fusion method, terms as pose dual-stream network (PDSN), is devised, aiming to test the feasibility of monitoring the training actions of rehabilitating stroke patients in care management. In particular, this deep-learning-based algorithm combines human pose estimation and dual-stream networks in an innovative way. We utilize an improved OpenPose to estimate human pose from videos obtained by the low-cost monocular camera. In dual-stream networks, the spatial and motion streams are flexibly integrated. The spatial stream network combines the Gated Recurrent Unit (GRU) and attention mechanism to extract spatiotemporal data, while the motion stream network is composed of improved multi-layer 1D Convolutional Neural Networks (CNN), which enhanced by causal and dilated convolution skillfully. Additionally, an adaptive weight fusion strategy is used to fuse the two networks for the final action classification. Results show high accuracy on two public datasets and a dataset created by us, which validate the superiority and feasibility of our method.
- 25 May 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 9, 78174-78187
Open AccessTesting the Feasibility of a Multi-Model Fusion Method for Monitoring the Action of Rehabilitating Stroke Patients in Care Management
Post-stroke care encounters challenges, including high cost, lack of professionals, and insufficient rehabilitation state evaluation. Computer technology can alleviate these issues, as it allows health care professionals (HCP) to quantify the workload and thus enhance rehabilitation care quality. In this paper, a novel multi-model fusion method, terms as pose dual-stream network (PDSN), is devised, aiming to test the feasibility of monitoring the training actions of rehabilitating stroke patients in care management. In particular, this deep-learning-based algorithm combines human pose estimation and dual-stream networks in an innovative way. We utilize an improved OpenPose to estimate human pose from videos obtained by the low-cost monocular camera. In dual-stream networks, the spatial and motion streams are flexibly integrated. The spatial stream network combines the Gated Recurrent Unit (GRU) and attention mechanism to extract spatiotemporal data, while the motion stream network is composed of improved multi-layer 1D Convolutional Neural Networks (CNN), which enhanced by causal and dilated convolution skillfully. Additionally, an adaptive weight fusion strategy is used to fuse the two networks for the final action classification. Results show high accuracy on two public datasets and a dataset created by us, which validate the superiority and feasibility of our method.