Dual-Space Masked Reconstruction for Robust Self-Supervised Human Activity Recognition
The DOI matches its Crossref record; 24 of the 24 references it cites were found in a registry or an index.
Abstract
Human Activity Recognition (HAR) based on wearable sensors faces critical challenges, including limited labeled data, distribution shifts, and sensitivity to sensor noise. To address these issues, this paper proposes a novel self-supervised learning (SSL) framework that leverages dual-space masked reconstruction to learn robust and generalizable representations for sensor-based HAR. Specifically, we design a brand-new pretext task in the pre-training stage, which learns representations by reconstructing the original signals with partial masks in both the original space and the latent space. For the reconstruction task in the original space, we adopt an approach similar to Masked Autoencoders (MAE). In the latent space, we introduce a Spectral block to extract more discriminative representations from the time-domain and frequency-domain information. Then, we implement feature-level reconstruction using a Mean Teacher Network. The feature extractor trained through this pretext task is subsequently utilized for downstream challenging HAR tasks. Experiments on four public datasets (MotionSense, UCI-HAR, PAMAP2 and RealWorld) demonstrate the framework's superiority over state-of-the-art SSL methods, achieving an average 2.45% improvement in Macro F1-score under fine-tuning protocols. This work achieves excellent activity recognition performance in labeled data scarcity and Non- IID data scenarios that are closer to the display world by unifying local and global signal interaction.…
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
Masked Autoencoders Are Scalable Vision Learners
He, K., Chen, X., Xie, S., Li, Y., Dollar, P., & Girshick, R. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.01553 · Cited by 8,489
Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition
Ordóñez, F., & Roggen, D. (2016) · Sensors · doi:10.3390/s16010115 · Cited by 2,761
Deep learning for sensor-based activity recognition: A survey
Wang, J., Chen, Y., Hao, S., Peng, X., & Hu, L. (2019) · Pattern Recognition Letters · doi:10.1016/j.patrec.2018.02.010 · Cited by 2,014 · cited 2×
“This instability arises because 𝐸𝑛𝑐𝑜𝑑𝑒𝑟 ′ and 𝐹 ′ are updated in each iteration, causing the target vector to vary across iterations and thus become unstable [33].”
Introducing a New Benchmarked Dataset for Activity Monitoring
Reiss, A., & Stricker, D. (2012) · 2012 16th International Symposium on Wearable Computers · doi:10.1109/iswc.2012.13 · Cited by 1,456
SimMIM: a Simple Framework for Masked Image Modeling
Xie, Z., Zhang, Z., Cao, Y., Lin, Y., Bao, J., Yao, Z., Dai, Q., & Hu, H. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.00943 · Cited by 1,255 · cited 2×
“In computer vision, influential masked modeling frameworks such as BEiT [1], simMIM [35], and Masked Autoen-coders (MAE) [9] have demonstrated exceptional capabilities in visual representation learning.”
On-body localization of wearable devices: An investigation of position-aware activity recognition
Sztyler, T., & Stuckenschmidt, H. (2016) · 2016 IEEE International Conference on Pervasive Computing and Communications (PerCom) · doi:10.1109/percom.2016.7456521 · Cited by 369 · cited 3×
“UCI-HAR[31] : Widely-used benchmark (2012 release) containing six daily activities recorded at 50Hz from 30 participants (18-49 years) using Samsung Galaxy S I sensors.”
Multi-task Self-Supervised Learning for Human Activity Detection
Saeed, A., Ozcelebi, T., & Lukkien, J. (2019) · Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies · doi:10.1145/3328932 · Cited by 346 · cited 3×
“First, collecting sensor signals from user devices and annotating them with labels may compromise the scalability of data collection [28], as manual annotation of raw sensor data constitutes a labor-intensive task requiring substantial temporal and financial resources.”
Protecting Sensory Data against Sensitive Inferences
Malekzadeh, M., Clegg, R. G., Cavallaro, A., & Haddadi, H. (2018) · Proceedings of the 1st Workshop on Privacy by Design in Distributed Systems · doi:10.1145/3195258.3195260 · Cited by 154 · cited 2×
“MotionSense [22]: Collected via smartphone accelerometer and gyroscope from 24 participants (14M/10F, 16-18 years) at 50Hz sampling rate, demonstrating high diversity.”
Contrastive Predictive Coding for Human Activity Recognition
Haresamudram, H., Essa, I., & Plötz, T. (2021) · Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies · doi:10.1145/3463506 · Cited by 150
Masked reconstruction based self-supervision for human activity recognition
Haresamudram, H., Beedu, A., Agrawal, V., Grady, P. L., Essa, I., Hoffman, J., & Plötz, T. (2020) · Proceedings of the 2020 International Symposium on Wearable Computers · doi:10.1145/3410531.3414306 · Cited by 146 · cited 2×
“We adopt the random masking strategy[6], with a masking ratio 𝑝 .”
Research works printed but not linked: 11
- Hangbo Bao, Li Dong, Songhao Piao, and Furu Wei. 2021. Beit: Bert pre-training of image transformers. arXiv preprint arXiv:2106.08254 (2021). · arXiv
- Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. 2020. A simple framework for contrastive learning of visual representations. In International conference on machine learning. PmLR, 1597-1607.
- Xiaobing Chen, Xiangwei Zhou, Mingxuan Sun, and Hao Wang. 2025. Temporal Contrastive Learning for Sensor-Based Human Activity Recognition: A Self-Supervised Approach. Sensors Journal, IEEE, Vol. 25, 1 (2025), 1839-1850.
- Jacob Devlin, Ming Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, vol. 6. long and short papers: Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL HLT 2019), 2-7 June 2019, Minneapolis, Minnesota, USA.
- Sannara Ek Riccardo Presotto Gabriele Civitarese François Portet Philippe Lalanda and Claudio Bettini. 2024. Comparing Self-Supervised Learning Techniques for Wearable Human Activity Recognition. arXiv:2404.15331 [eess.SP] https://arxiv.org/abs/2404.15331 · arXiv
- 10.1109/cvpr52688.2022.01553 · DOI · Paper page
- Md.Milon Islam, Sheikh Nooruddin, Fakhri Karray, and Ghulam Muhammad. 2022. Human Activity Recognition Using Tools of Convolutional Neural Networks: A State of the Art Review, Data Sets, Challenges and Future Prospects. Computers in Biology and Medicine (2022).
- Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, and Radu Soricut. 2019. Albert: A lite bert for self-supervised learning of language representations. arXiv preprint arXiv:1909.11942 (2019). · arXiv
- Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692 (2019). · arXiv
- Badri N. Patro Vinay P. Namboodiri and Vijay Srinivas Agneeswaran. 2023. SpectFormer: Frequency and Attention is what you need in a Vision Transformer. arXiv:2304.06446 [cs.CV] https://arxiv.org/abs/2304.06446 · arXiv
- Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. 2019. DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter. arXiv preprint arXiv:1910.01108 (2019). · arXiv
Cited by
A Survey on Multimodal Wearable Sensor-based Human Action Recognition
Ni, J., Guo, B., Haque, S. T., Ngu, A. H. H., & Ding, C. (2026) · IEEE Internet of Things Journal · doi:10.1109/jiot.2026.3716688 · Cited by 17
“…SSL methods range from pseudo-labeling and masked reconstruction to contrastive, temporal-consistency, and hierarchical multiview learning, enabling models to exploit unlabeled data while preserving spatial–temporal and cross-modal structure [163], [164], [165], [166], [167], [168], [169], [170].”
Metadata
- Authors
- Xiao, S., Deng, J., Tang, C., & Huang, Z.
- Affiliations
- China University of Mining and Technology (CN)
- Venue
- Proceedings of the 34th ACM International Conference on Information and Knowledge Management
- Publisher
- ACM
- Published
- 2025-11-10
- Pages
- 3488-3497
- ISSN
- 2155-0751 (ISSN-L)
- OpenAlex
- W4416016023
- Semantic Scholar
- b427e8531109745b923f6faadf796953bfd25775
- Cited by
- 0
- References
- 24
- FWCI
- 0.00
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- ACM International Conference on Information and Knowledge Management (CIKM) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Human Pose and Action Recognition
- Physical Sciences › Engineering › Biomedical Engineering › Gait Recognition and Analysis
- Keywords
- human activity recognition, self-supervised learning, masked reconstruction, robustness
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
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Dual-Space Masked Reconstruction for Robust Self-Supervised Human Activity Recognition. https://app.cochat.ai/papers/10.1145/3746252.3761215
- The paper
- Xiao, S., Deng, J., Tang, C., & Huang, Z. (2025). Dual-Space Masked Reconstruction for Robust Self-Supervised Human Activity Recognition. Proceedings of the 34th ACM International Conference on Information and Knowledge Management. https://doi.org/10.1145/3746252.3761215
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.