A Weight-Aware-Based Multisource Unsupervised Domain Adaptation Method for Human Motion Intention Recognition
Open accessThe DOI matches its Crossref record; 64 of the 64 references it cites were found in a registry or an index.
Abstract
Accurate recognition of human motion intention (HMI) is beneficial for exoskeleton robots to improve the wearing comfort level and achieve natural human-robot interaction. A classifier trained on labeled source subjects (domains) performs poorly on unlabeled target subject since the difference in individual motor characteristics. The unsupervised domain adaptation (UDA) method has become an effective way to this problem. However, the labeled data are collected from multiple source subjects that might be different not only from the target subject but also from each other. The current UDA methods for HMI recognition ignore the difference between each source subject, which reduces the classification accuracy. Therefore, this article considers the differences between source subjects and develops a novel theory and algorithm for UDA to recognize HMI, where the margin disparity discrepancy (MDD) is extended to multisource UDA theory and a novel weight-aware-based multisource UDA algorithm (WMDD) is proposed. The source domain weight, which can be adjusted adaptively by the MDD between each source subject and target subject, is incorporated into UDA to measure the differences between source subjects. The developed multisource UDA theory is theoretical and the generalization error on target subject is guaranteed. The theory can be transformed into an optimization problem for UDA, successfully bridging the gap between theory and algorithm. Moreover, a lightweight network is employed to guarantee the real-time of classification and the adversarial learning between feature generator and ensemble classifiers is utilized to further improve the generalization ability. The extensive experiments verify theoretical analysis and show that WMDD outperforms previous UDA methods on HMI recognition tasks.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
Generative adversarial networks
Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., & Bengio, Y. (2020) · Communications of the ACM · doi:10.1145/3422622 · Cited by 18,384
Domain-Adversarial Training of Neural Networks
Ganin, Y., Ustinova, E., Ajakan, H., Germain, P., Larochelle, H., Laviolette, F., Marchand, M., & Lempitsky, V. (2017) · Advances in Computer Vision and Pattern Recognition · doi:10.1007/978-3-319-58347-1_10 · Cited by 8,318 · cited 2×
“To alleviate the performance reduction caused by this discrepancy, single-source UDA (SUDA) utilizes the adversarial-based methods to align two domains [11]–[13] or explores different metric learning schemes to minimize this divergence [3], [25], [32].”
Learning Transferable Features with Deep Adaptation Networks
Long, M., Cao, Y., Wang, J., & Jordan, M. I. (2015) · arXiv · doi:10.48550/arxiv.1502.02791 · Cited by 4,505 · cited 1×
“For challenge 1, Jensen Shannon Divergence [26], Maximum Mean Discrepancy [27] and Wasserstein Distance [28] are widely employed to measure the discrepancy between source and target domains in SUDA.”
A theory of learning from different domains
Ben-David, S., Blitzer, J., Crammer, K., Kulesza, A., Pereira, F., & Vaughan, J. W. (2010) · Machine Learning · doi:10.1007/s10994-009-5152-4 · Cited by 3,515 · cited 4×
“Ben-David et al. [1], [31] conducted the pioneering theoretical works in domain adaptation field.”
InfluentialEnsemble Methods: Foundations and Algorithms
Zhou, Z.-H. (2012) · doi:10.1201/b12207 · Cited by 2,534
Unsupervised Domain Adaptation by Backpropagation
Ganin, Y., & Lempitsky, V. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1409.7495 · Cited by 2,521 · cited 1×
“For challenge 1, Jensen Shannon Divergence [26], Maximum Mean Discrepancy [27] and Wasserstein Distance [28] are widely employed to measure the discrepancy between source and target domains in SUDA.”
Maximum Classifier Discrepancy for Unsupervised Domain Adaptation
Saito, K., Watanabe, K., Ushiku, Y., & Harada, T. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00392 · Cited by 2,161
Analysis of Representations for Domain Adaptation
Ben-David, S., Blitzer, J., Crammer, K., & Pereira, F. (2007) · Advances in Neural Information Processing Systems 19 · doi:10.7551/mitpress/7503.003.0022 · Cited by 2,019 · cited 2×
“Ben-David et al. [1], [31] conducted the pioneering theoretical works in domain adaptation field.”
Foundations of Machine Learning
Mohri, M., Rostamizadeh, A., & Talwalkar, A. (2018) · doi:10.7551/mitpress/10290.001.0001 · Cited by 1,122
Mind the Class Weight Bias: Weighted Maximum Mean Discrepancy for Unsupervised Domain Adaptation
Yan, H., Ding, Y., Li, P., Wang, Q., Xu, Y., & Zuo, W. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.107 · Cited by 613 · cited 2×
“For the second-type methods, Yan et al. [61] calculated domain weight based on class prior distribution, and proposed weighted domain adaption network to alleviate the effect of class weight bias.”
Domain Adaptation: Learning Bounds and Algorithms
Mansour, Y., Mohri, M., & Rostamizadeh, A. (2009) · arXiv · doi:10.48550/arxiv.0902.3430 · Cited by 504 · cited 2×
“Mansour et al. [41] extended the zero-one loss of [31] to the general loss function of binary classification and developed a generalization theory.”
Joint Distribution Optimal Transportation for Domain AdaptationPreprint
Courty, N., Flamary, R., Habrard, A., & Rakotomamonjy, A. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1705.08848 · Cited by 409 · cited 1×
“For challenge 1, Jensen Shannon Divergence [26], Maximum Mean Discrepancy [27] and Wasserstein Distance [28] are widely employed to measure the discrepancy between source and target domains in SUDA.”
Learning Bounds for Domain Adaptation
Blitzer, J. C., Crammer, K., Kulesza, A., Pereira, F. M. B., & Wortman, J. R. (2007) · Cited by 393
Barshan, B., & Yuksek, M. C. (2014) · The Computer Journal · doi:10.1093/comjnl/bxt075 · Cited by 346 · cited 2×
“The other is the daily and sports activities data set (DSADS) collected by Bilken University [10].”
Advances and Challenges in Meta-Learning: A Technical Review
Vettoruzzo, A., Bouguelia, M.-R., Vanschoren, J., Rögnvaldsson, T., & Santosh, K. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3357847 · Cited by 314 · cited 1×
“However, the performance of these methods would be limited if data in source domain is insufficient [51].”
Unsupervised Domain Adaptation of Object Detectors: A Survey
Oza, P., Sindagi, V. A., VS, V., & Patel, V. M. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3217046 · Cited by 313 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Bridging Theory and Algorithm for Domain Adaptation
Zhang, Y., Liu, T., Long, M., & Jordan, M. I. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.05801 · Cited by 289 · cited 10×
“Tosolve the above problems, motivated by the margin disparity discrepancy [3] and distribution-weighted hypothesis [29], we extend MDD in SUDA that contains more generalization information to MUDA field to accurately measure the discrepancy between source and target domains, and propose a novel…”
InfluentialDynamic Weighted Learning for Unsupervised Domain Adaptation
Xiao, N., & Zhang, L. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.01499 · Cited by 181
Learning Multiple Tasks with Multilinear Relationship Networks
Long, M., Cao, Z., Wang, J., & Yu, P. S. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1506.02117 · Cited by 173
Bayesian Estimation of Human Impedance and Motion Intention for Human–Robot Collaboration
Yu, X., He, W., Li, Y., Xue, C., Li, J., Zou, J., & Yang, C. (2021) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2019.2940276 · Cited by 162 · cited 1×
“To improve the wearing comfort level and achieve natural human-robot interaction, human motion intention (HMI) has gained great concerns in exoskeleton robots and human-robot interaction [2], [23], [47].”
Robust Real-Time Musculoskeletal Modeling Driven by Electromyograms
Durandau, G., Farina, D., & Sartori, M. (2018) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2017.2704085 · Cited by 157 · cited 2×
“The model-based methods involve the kinematics model [19], the dynamic model [17], and the musculoskeletal model [18].”
Estimating human joint moments unifies exoskeleton control, reducing user effort
Molinaro, D. D., Kang, I., & Young, A. J. (2024) · Science Robotics · doi:10.1126/scirobotics.adi8852 · Cited by 122 · cited 1×
“To improve the wearing comfort level and achieve natural human-robot interaction, human motion intention (HMI) has gained great concerns in exoskeleton robots and human-robot interaction [2], [23], [47].”
Unsupervised Multi-Class Domain Adaptation: Theory, Algorithms, and Practice
Zhang, Y., Deng, B., Tang, H., Zhang, L., & Jia, K. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2020.3036956 · Cited by 116
Attention-Based Multi-Source Domain Adaptation
Zuo, Y., Yao, H., & Xu, C. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3065254 · Cited by 95 · cited 4×
“The source domain weight was used to integrate source domain differences, and mainly calculated through solving related optimization problem [33], [35] or calculating the discrepancy or similarity between source and target domains [34], [59]–[62].”
InfluentialHu, B. H., Rouse, E. J., & Hargrove, L. J. (2018) · Frontiers in Robotics and AI · doi:10.3389/frobt.2018.00014 · Cited by 89 · cited 2×
“For the ENABL3S dataset [9], ten subjects are invited to walk on several terrains and switch locomotion modes between sitting, standing, level ground walking, stair ascent, stair descent, ramp ascent, and ramp descent.”
王, J. 璟. W., Chen, J., Lin, J., Sigal, L., & de Silva, C. W. (2021) · Pattern Recognition · doi:10.1016/j.patcog.2021.107943 · Cited by 86
Finn, C., & Levine, S. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1710.11622 · Cited by 86 · cited 1×
“Meta learning, such as MAML [50], acquires knowledge from multiple tasks, enabling faster adaptation and generalization to new tasks.”
Cross-Subject and Cross-Modal Transfer for Generalized Abnormal Gait Pattern Recognition
Gu, X., Guo, Y., Deligianni, F., Lo, B. P. L., & Yang, G. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.3009448 · Cited by 83 · cited 1×
“Some methods make improvements to the perception network framework itself, such as using multi-encoder autoencoder [65], to improve performance across subjects.”
Ren, C., Jiang, B., Lu, N., Simani, S., & Gao, F. (2023) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2023.3338768 · Cited by 78 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Qiu, S., Guo, W. P., Caldwell, D. G., & Chen, F. (2020) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2020.2968845 · Cited by 77 · cited 1×
“The model-based methods involve the kinematics model [19], the dynamic model [17], and the musculoskeletal model [18].”
Class-aware sample reweighting optimal transport for multi-source domain adaptation
Wang, S., Wang, B., Zhang, Z., Heidari, A. A., & Chen, H. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.12.048 · Cited by 77 · cited 1×
“The second type may not better capture the discrepancy or similarity relationship between domains [24] when source and target distributions are complex.”
Unsupervised Domain Adaptation Based on Source-Guided Discrepancy
Kuroki, S., Charoenphakdee, N., Bao, H., Honda, J., Sato, I., & Sugiyama, M. (2019) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v33i01.33014122 · Cited by 74
Human Lower Limb Motion Intention Recognition for Exoskeletons: A Review
Li, L., Cao, G., Liang, H., Zhang, Y.-P., & Cui, F. (2023) · IEEE Sensors Journal · doi:10.1109/jsen.2023.3328615 · Cited by 71 · cited 2×
“Due to the difference in individual motor characteristics, such as kinematic properties, most current methods require labeling a large amount of data and training specific classifiers for each individual, which is burdensome [16].”
Active Human-Following Control of an Exoskeleton Robot With Body Weight Support
Li, G., Li, Z., Su, C., & Tian, X. (2023) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2023.3253181 · Cited by 64 · cited 1×
“Ex-oskeleton robots, including rehabilitation robots [42], [44], [46] and power-assisted robots [43], [45], have become one of the important ways to treat and recover these diseases.”
Self-citationElectromyography-Driven Progressive Assist-as-Needed Control for Lower Limb Exoskeleton
Gui, K., Tan, U.-X., Liu, H., & Zhang, D. (2020) · IEEE Transactions on Medical Robotics and Bionics · doi:10.1109/tmrb.2020.2970222 · Cited by 63 · cited 1×
“The model-based methods involve the kinematics model [19], the dynamic model [17], and the musculoskeletal model [18].”
New Analysis and Algorithm for Learning with Drifting Distributions
Mohri, M., & Medina, A. M. (2012) · Lecture notes in computer science · doi:10.1007/978-3-642-34106-9_13 · Cited by 60 · cited 1×
“First, Rademacher complexity that is widely used in the generalization theory is introduced to measure the richness of a certain hypothesis space [3], [39].”
Domain Aggregation Networks for Multi-Source Domain AdaptationPreprint
Wen, J., Greiner, R., & Schuurmans, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.05352 · Cited by 59 · cited 6×
“A lightweight network is employed to guarantee the real-time of classification, instead of using multiple complex networks [33] that is difficult to guarantee the real-time.”
InfluentialDevelopment of an Untethered Adaptive Thumb Exoskeleton for Delicate Rehabilitation Assistance
Li, G., Cheng, L., Gao, Z., Xia, X., & Jiang, J. (2022) · IEEE Transactions on Robotics · doi:10.1109/tro.2022.3180832 · Cited by 58 · cited 1×
“Ex-oskeleton robots, including rehabilitation robots [42], [44], [46] and power-assisted robots [43], [45], have become one of the important ways to treat and recover these diseases.”
Self-citationContinuous Estimation of Human Knee Joint Angles by Fusing Kinematic and Myoelectric Signals
Sun, N., Cao, M., Chen, Y., Chen, Y., Wang, J., Wang, Q., Chen, X., & Liu, T. (2022) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2022.3200485 · Cited by 56 · cited 1×
“Model-free methods based on deep networks, which can obtain higher-level features from sensor signals without domain-specific knowledge, have achieved great success in recent years, such as CNN [38], LSTM [37], CNN-BiLSTM [36].”
Unsupervised Cross-Subject Adaptation for Predicting Human Locomotion Intent
Zhang, K., Wang, J., de Silva, C. W., & Fu, C. (2020) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2020.2966749 · Cited by 53 · cited 3×
“Unsupervised domain adaptation (UDA) is a common method to solve the above problem [6]–[8], [48], where the source subjects (domains) data are labeled and the target subject (domain) data are unlabeled.”
Bruinsma, J., & Carloni, R. (2021) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2021.3086843 · Cited by 49 · cited 1×
“Model-free methods based on deep networks, which can obtain higher-level features from sensor signals without domain-specific knowledge, have achieved great success in recent years, such as CNN [38], LSTM [37], CNN-BiLSTM [36].”
Multisource Heterogeneous Domain Adaptation With Conditional Weighting Adversarial Network
Yao, Y., Li, X., Zhang, Y., & Ye, Y. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2021.3105868 · Cited by 47 · cited 5×
“The source domain weight was used to integrate source domain differences, and mainly calculated through solving related optimization problem [33], [35] or calculating the discrepancy or similarity between source and target domains [34], [59]–[62].”
InfluentialHu, R., Chen, L., Miao, S., & Tang, X. (2023) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v37i5.25743 · Cited by 44
Mo, Z., Zhang, Z., Miao, Q., & Tsui, K. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2022.3223783 · Cited by 43 · cited 1×
“Unsupervised domain adaptation (UDA) is a common method to solve the above problem [6]–[8], [48], where the source subjects (domains) data are labeled and the target subject (domain) data are unlabeled.”
Multidomain Adaptation With Sample and Source Distillation
Li, K., Lü, J., Zuo, H., & Zhang, G. (2023) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2023.3236008 · Cited by 40 · cited 1×
“However, the above UDA methods for HMI recognition don’t consider the difference between the source subjects, which may result in a sub-optimal solution to the problem and hinder the improvement of classifier performance [49].”
Wang, C., Wang, Z., Liu, H., Dong, H., & Lü, G. (2024) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2024.3385533 · Cited by 33 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Li, Z., Zhang, T., Huang, P., & Li, G. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2022.3211925 · Cited by 32 · cited 1×
“Ex-oskeleton robots, including rehabilitation robots [42], [44], [46] and power-assisted robots [43], [45], have become one of the important ways to treat and recover these diseases.”
Self-citationSA-SVM-Based Locomotion Pattern Recognition for Exoskeleton Robot
Yin, Z., Zheng, J., Huang, L., Gao, Y., Peng, H., & Yin, L. (2021) · Applied Sciences · doi:10.3390/app11125573 · Cited by 31
Yang, J., & Yin, Y. (2020) · IEEE Sensors Journal · doi:10.1109/jsen.2020.3019053 · Cited by 30 · cited 1×
“Model-free methods based on deep networks, which can obtain higher-level features from sensor signals without domain-specific knowledge, have achieved great success in recent years, such as CNN [38], LSTM [37], CNN-BiLSTM [36].”
Development and Validation of a Self-Aligning Knee Exoskeleton With Hip Rotation Capability
Li, G., Liang, X., Lu, H., Su, T., & Hou, Z. (2024) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2024.3354806 · Cited by 27 · cited 1×
“Ex-oskeleton robots, including rehabilitation robots [42], [44], [46] and power-assisted robots [43], [45], have become one of the important ways to treat and recover these diseases.”
Self-citationLiu, J., Zhou, X., He, B., Li, P., Wang, C., & Wu, X. (2022) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2022.3185380 · Cited by 26 · cited 1×
“The model-free methods map the sensing signals into target HMI directly without parameter identification [20]–[22].”
Self-citationMulti-Source Domain Adaptation with Mixture of Joint Distributions
Chen, S. (2024) · Pattern Recognition · doi:10.1016/j.patcog.2024.110295 · Cited by 25 · cited 7×
“Compared with the domain weight optimized constantly by solving unconstrained problem [35], the proposed method can reduce the computation cost and enhance algorithm stability.”
InfluentialWang, C., Guo, Z., Duan, S., He, B., Yuan, Y., & Wu, X. (2021) · Frontiers in Neuroscience · doi:10.3389/fnins.2021.645374 · Cited by 20 · cited 1×
“The model-free methods map the sensing signals into target HMI directly without parameter identification [20]–[22].”
Huang, P., Li, Z., Zhou, M., & Kan, Z. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2022.3222564 · Cited by 18 · cited 1×
“Ex-oskeleton robots, including rehabilitation robots [42], [44], [46] and power-assisted robots [43], [45], have become one of the important ways to treat and recover these diseases.”
Li, M., Zhang, C., Zhao, W., & Zhou, W. (2024) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2024.3399741 · Cited by 16 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Gaussian-guided feature alignment for unsupervised cross-subject adaptation
Zhang, K., Chen, J., Wang, J., Leng, Y., de Silva, C. W., & Fu, C. (2021) · Pattern Recognition · doi:10.1016/j.patcog.2021.108332 · Cited by 14 · cited 2×
“Zhang et al. [7] proposed a novel non-adversarial cross-subject adaptation method (EDHKD) for HMI recognition.”
PnP-GA+: Plug-and-Play Domain Adaptation for Gaze Estimation Using Model Variants
Liu, R., Liu, Y., Wang, H., & Lu, F. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3348528 · Cited by 14 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Hao, X., Li, Z., Huang, P., Shi, P. J., & Li, G. (2023) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2023.3345919 · Cited by 13 · cited 1×
“To improve the wearing comfort level and achieve natural human-robot interaction, human motion intention (HMI) has gained great concerns in exoskeleton robots and human-robot interaction [2], [23], [47].”
Self-citationFew-Shot Unsupervised Domain Adaptation via Meta Learning
Yang, W., Yang, C., Huang, S., Wang, L., & Yang, M. (2022) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme52920.2022.9859804 · Cited by 12 · cited 1×
“Some meta learning techniques [52] can be combined with domain adaptation to solve the above problem.”
Ren, C., Jiang, B., & Lu, N. (2024) · IEEE Transactions on Reliability · doi:10.1109/tr.2024.3427813 · Cited by 12 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Maximum likelihood weight estimation for partial domain adaptation
Wen, L., Chen, S., Hong, Z., & Zheng, L. (2024) · Information Sciences · doi:10.1016/j.ins.2024.120800 · Cited by 11 · cited 6×
“The source domain weight was used to integrate source domain differences, and mainly calculated through solving related optimization problem [33], [35] or calculating the discrepancy or similarity between source and target domains [34], [59]–[62].”
InfluentialA Novel Angular-Based Unsupervised Domain Adaptation Framework for Image Classification
Mishra, S., & Sanodiya, R. K. (2023) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2023.3293077 · Cited by 9 · cited 1×
“Currently, the UDA methods have also been widely applied in many other fields, such as fault diagnosis [55], [63], physical information estimation [53], classification tasks [56], [57], and object identification [54], [64].”
Ensemble diverse hypotheses and knowledge distillation for unsupervised cross-subject adaptation
Zhang, K., Chen, J., Wang, J., Chen, X., Leng, Y., de Silva, C. W., & Fu, C. (2022) · Information Fusion · doi:10.1016/j.inffus.2022.12.023 · Cited by 8
Mixture Weight Estimation and Model Prediction in Multi-source Multi-target Domain Adaptation
Deng, Y., Kuzborskij, I., & Mahdavi, M. (2023) · neural information processing systems · doi:10.52202/075280-0216 · Cited by 3
Research works printed but not linked: 2
- 10.5555/2969033.2969125 · doi:10.5555/2969033.2969125
- Van der Maaten, 2008, Visualizing data using t-SNE, J. Mach. Learn. Res., 9, 2579
Cited by
Fuzzy Sliding Mode Control for Dynamic Walking Assistance in Lower Limb Exoskeletons
Gebrecherkos, T. H., Berhe, A. G., Gurusamy, S., & Abdissa, C. M. (2025) · IEEE Access · doi:10.1109/access.2025.3607679 · Cited by 29
“Other neurological conditions, such as stroke and multiple sclerosis, can also impair the nervous system’s ability to control lower limb movements, leading to varying degrees of disability [6], [7], [8].”
An enhanced ResNet deep learning method for multimodal signal-based locomotion intention recognition
Sun, H., Gu, X., Zhang, Y., Sun, F., Zhang, S., Wang, D., & Yu, H. (2024) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.107254 · Cited by 12
Wang, Z., & Kang, K. (2025) · Scientific Reports · doi:10.1038/s41598-025-18444-6 · Cited by 5
Computer vision-based laser communication system for robust optical beam tracking and alignment
Li, S. (2025) · Scientific Reports · doi:10.1038/s41598-025-17695-7 · Cited by 5
Contrastive domain adaptation with time-lag attention for wearable sensor time-series
Zhang, G., Zhang, H., Jia, M., & Liu, Z. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.131087 · Cited by 4
Metadata
- Authors
- Liu, X.-Y., Li, G., Zhou, X.-H., Liang, X., & Hou, Z.-G.
- Affiliations
- Chinese Academy of Sciences (CN)
- Venue
- IEEE Transactions on Cybernetics
- Published
- 2025-07
- Volume
- 55
- Issue
- 7
- Pages
- 3131-3143
- ISSN
- 2168-2267 (ISSN-L)
- 2168-2275
- OpenAlex
- W4410536891
- Cited by
- 36
- References
- 63
- Citations by year
- 2026: 18 · 2025: 17
- FWCI
- 13.99
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Cybernetics (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- National Natural Science Foundation of China (62473365, U22A2056, 62373013)
- Beijing Natural Science Foundation (L222053, L232021, L242101)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Engineering › Biomedical Engineering › Muscle activation and electromyography studies
- Physical Sciences › Engineering › Biomedical Engineering › Prosthetics and Rehabilitation Robotics
- Keywords
- unsupervised domain adaptation, robotic exoskeleton, human-robot interaction, adversarial learning, lightweight neural networks, real-time classification, generalization error
- MeSH terms
- Unsupervised Machine Learning, Algorithms, Humans, Movement, Pattern Recognition, Automated, Intention
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.