The DOI matches its Crossref record; 56 of the 57 references it cites were found in a registry or an index; 1 of the 30 sources its claims cite were checked for support.
Comparing self-supervised learning techniques for wearable human activity recognition
Open accessAbstract
Human Activity Recognition (HAR) based on the sensors of mobile/wearable devices aims to detect the physical activities performed by humans in their daily lives. Although supervised learning methods are the most effective in this task, their effectiveness is constrained to using a large amount of labeled data during training. While collecting raw unlabeled data can be relatively easy, annotating data is challenging due to costs, intrusiveness, and time constraints. To address these challenges, this paper explores alternative approaches for accurate HAR using a limited amount of labeled data. In particular, we have adapted recent Self-Supervised Learning (SSL) algorithms to the HAR domain and compared their effectiveness. We investigate three state-of-the-art SSL techniques of different families: contrastive, generative, and predictive. Additionally, we evaluate the impact of the underlying neural network on the recognition rate by comparing state-of-the-art CNN and transformer architectures. Our results show that a Masked Auto Encoder approach significantly outperforms other SSL approaches, including SimCLR, commonly considered one of the best-performing SSL methods in the HAR domain. The code and the pre-trained SSL models are publicly available for further research and development.
Overview
CoChat Summary
## TL;DR
The paper compares three SSL approaches for wearable HAR—SimCLR, MAE, and data2vec—across CNN (iSPLInception) and transformer (HART) backbones using a Leave-One-Dataset-Out benchmark. MAE is the most robust method, especially in the frozen-feature-extractor setting and in low-label regimes, while SimCLR performs worst overall.
## Key findings
- With the feature extractor frozen, MAE outperformed the other SSL methods and the supervised baseline on average: 80.90 ± 10.88 (iSPL) and 82.56 ± 11.38 (HART), versus supervised 65.74 ± 16.56 and 74.95 ± 15.92.
- In the frozen, dataset-level results, MAE was best on every dataset reported; for example, on HHAR it reached 66.11 ± 1.05 (iSPL) and 66.64 ± 1.61 (HART), and on UCI 92.17 ± 0.83 (iSPL) and 93.40 ± 0.38 (HART).
- In the unfrozen setting with full training data, there was no clear winner; performances were very close, with all methods reaching about the same recognition rates for a given architecture.
- In scarce-label experiments, MAE was consistently strongest overall and achieved above 60% F1 with only 10 samples per class in the frozen setting; SimCLR performed very poorly, especially on RealWorld.
## Methods
- The study adapts and compares three SSL families for HAR: SimCLR (contrastive), MAE (generative/masking), and data2vec (predictive), each paired with two architectures: iSPLInception and HART.
- Evaluation uses Leave-One-Dataset-Out (LODO) across six public inertial-sensor datasets: HHAR, MobiAct, MotionSense, RealWorld, UCI, and PAMAP2; data were downsampled to 50 Hz, windowed into 128-sample segments with 50% overlap, yielding 147 subjects, 10 activities, 213,289 samples (about 151 hours), and 12 devices.
- Main analysis: pretraining for 300 epochs with batch size 128, then fine-tuning for 100 epochs with batch size 64; performance is reported using macro F1, averaged over 5 random runs, under both frozen and unfrozen feature-extractor settings and across label-scarcity scenarios from 1 to 100 samples per class.
## Limitations
- The paper states that the study was limited to three SSL techniques and two models, and that it was conducted only as offline experiments; behavior in real-world settings remains to be evaluated.
- The authors note that the available HAR datasets are still much smaller and less diverse than those in vision or text, even though LODO helps mitigate limited data.
- Evident from the text: the label-scarcity experiments are reported only on RealWorld and MotionSense for presentation purposes, even though complete results exist in the repository.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
1 Valid · 3 Not valid · 17 Not checked · 75 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“The majority of these works have focused on contrastive approaches [23, 24, 25, 26, 27].”
What the sources say
- Partly supportedcites [23] Haresamudram, H., Essa, I., Plötz, T.: Investigating enhancements to contrastive predictive coding for human activity recognition. In: 2023 IEEE International Conference on Pervasive Computing and Communications (PerCom), pp. 232–241 (2023). IEEE sourceExperimental — accuracy not yet measured
Our copy's Related Work (II) and Introduction (I) catalogue the SSL HAR literature and much of it is contrastive — SimCLR [24], [26], CPC [21], CPC-based change point detection [25], and the multi-sensor ColloSSL [35] / COCOA [36] — which lends direction to the claim. However, no passage states or implies that the majority of SSL work in HAR is contrastive: the same sections give equal standing to non-contrastive approaches (Multi-task self-supervision [19], Masked Reconstruction [20], Autoencoders [23], [29]) and explicitly note that SimSiam and BYOL 'are also siamese methods, albeit do not contrast the two views' (II). The Introduction's own summary list of methods for wearables contains only two contrastive entries out of four. So the source supports 'contrastive approaches are prominent in SSL for HAR' but not the quantitative 'majority' assertion attributed to it. (LLM, conf=0.68)
- Not checked yetcites [24] He, K., Chen, X., Xie, S., Li, Y., Dollár, P., Girshick, R.: Masked autoencoders are scalable vision learners. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 16000–16009 (2022) source
- Not checked yetcites [25] Hernandez, N., Lundström, J., Favela, J., McChesney, I., Arnrich, B.: Literature review on transfer learning for human activity recognition using mobile and wearable devices with environmental technology. SN Comput. Sci. 1, 1–16 (2020) source
- Not checked yetcites [26] Ignatov, A.: Real-time human activity recognition from accelerometer data using convolutional neural networks. Appl. Soft Comput. 62 (2018) source
- Not checked yetcites [27] Jain, Y., Tang, C.I., Min, C., Kawsar, F., Mathur, A.: Collossl: Collaborative self-supervised learning for human activity recognition. In: Proc. of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies 6(1) (2022) source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The findings indicate that MAE emerges as the most robust technique for the domain, surpassing all other approaches, including supervised pre-training methods.”
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our results show that a Masked Auto Encoder (MAE) approach significantly outperforms other SSL approaches, including SimCLR, commonly considered one of the best-performing SSL methods in the HAR domain.”
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Differences in performances are generally below a 1% margin between the best and performing techniques on both architectures.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In the data-scarce environment, we saw that self-supervised pre-training allowed the model to perform better in extremely low sample ranges when compared to starting from random, and results converged to equivalent performance with more training data.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
50 more claims
Create a free account to see all 96 claims and to verify the 29 references still open.
- It may state two separate claims as one. 19
- Probably not a substantial claim. 25
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“In the following, we describe the six datasets (HHAR, MobiAct, MotionSense, Realworld, UCI, and PAMAP2) employed in this study.”
“We gathered seven datasets, each possessing unique traits that distinguish them from one another.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“During the fine-tuning phase, we performed 100 epochs with a batch size of 64 on the left-out datasets.”
“the training process is constrained (from 200 epochs to only 50 epochs).”
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 · cited 6×
“However, the complete results on all seven datasets are also given in the shared code repository.”
InfluentialBootstrap Your Own Latent: A New Approach to Self-Supervised Learning
Grill, J.-B., Strub, F., Altché, F., Tallec, C., Richemond, P. H., Buchatskaya, E., Doersch, C., Pires, B. A., Guo, Z. D., Azar, M. G., et al. (2020) · arXiv · doi:10.48550/arxiv.2006.07733 · Cited by 3,846
Exploring Simple Siamese Representation Learning
Chen, X., & He, K. (2021) · 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr46437.2021.01549 · Cited by 3,725 · cited 1×
“Furthermore, unlike other siamese methods, such as BYOL (Grill et al. 2020) or SimSiam (Chen and He 2021), which employ additional components such as projection heads in the student model to mitigate collapse, Data2Vec does not utilize such mechanisms, making it more susceptible to this issue.”
On the Opportunities and Risks of Foundation Models
Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., et al. (2021) · arXiv · doi:10.48550/arxiv.2108.07258 · Cited by 2,676
A survey on semi-supervised learning
van Engelen, J. E., & Hoos, H. H. (2020) · Machine Learning · doi:10.1007/s10994-019-05855-6 · Cited by 2,666 · cited 1×
“For this purpose, Self-Supervised Learning (SSL) meth-ods (Van Engelen and Hoos 2020) can prove to be highly effective.”
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 1×
“In this case, the most effective approaches rely on deep learning, where high recognition rates are achieved through the use of supervised learning techniques (Wang et al. 2019).”
Mamba: Linear-Time Sequence Modeling with Selective State Spaces
Gu, A., & Dao, T. (2023) · arXiv · doi:10.48550/arxiv.2312.00752 · Cited by 1,579
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 · cited 4×
“This section provides an overview of the datasets used and the evaluation strategies employed to assess the performance of the pre-trained models obtained through various SSL techniques and architectures.”
InfluentialDoherty, A., Jackson, D., Hammerla, N., Plötz, T., Olivier, P., Granat, M. H., White, T., van Hees, V. T., Trenell, M. I., Owen, C. G., et al. (2017) · PLOS ONE · doi:10.1371/journal.pone.0169649 · Cited by 1,250 · cited 2×
“Despite recent advancements (Chan Chang et al. 2021; Doherty et al. 2017), there have yet to be publicly available Gold-standard datasets that are expansive and inclusive of diverse devices, positions, users, and sensors.”
Sensor-Based Activity Recognition
Chen, L., Hoey, J., Nugent, C. D., Cook, D. J., & Yu, Z. (2012) · IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) · doi:10.1109/tsmcc.2012.2198883 · Cited by 1,074 · cited 1×
“Human activity recognition (HAR) aims to recognize human actions automatically using sensor data (Chen et al. 2012).”
Research works printed but not linked: 8
- Anguita, D., Ghio, A., Oneto, L., Parra, X., Reyes-Ortiz, J.L.: A public domain dataset for human activity recognition using smartphones. In: 21st European Symposium on Artificial Neural Networks, ESANN 2013, Bruges, Belgium, April 24–26, 2013 (2013)
- Baevski, A., Hsu, W.-N., Xu, Q., Babu, A., Gu, J., Auli, M.: Data2vec: A general framework for self-supervised learning in speech, vision and language. In: Int. Conference on Machine Learning, pp. 1298–1312 (2022). PMLR
- Chen, T., Kornblith, S., Norouzi, M., Hinton, G.: A simple framework for contrastive learning of visual representations. In: International Conference on Machine Learning, pp. 1597–1607 (2020). PMLR
- Devlin, J., Chang, M.-W., Lee, K., Toutanova, K.: Bert: Pre-training of deep bidirectional transformers for language understanding. In: North American Chapter of the Association for Computational Linguistics (2019). https://api.semanticscholar.org/CorpusID:52967399
- Kalantidis, Y., Sariyildiz, M.B., Pion, N., Weinzaepfel, P., Larlus, D.: Hard negative mixing for contrastive learning. Adv. Neural. Inf. Process. Syst. 33, 21798–21809 (2020)
- Miao, S., Chen, L., Hu, R.: Spatial-temporal masked autoencoder for multi-device wearable human activity recognition. Proc. ACM Interact. Mob. Wear. Ubiq. Technol. 7(4), 1–25 (2024)
- Qin, X., Chen, Y., Wang, J., Yu, C.: Cross-dataset activity recognition via adaptive spatial-temporal transfer learning. ACM IMWUT 3(4), 1–25 (2019)
- Zhang, X., Zhao, Z., Tsiligkaridis, T., Zitnik, M.: Self-supervised contrastive pre-training for time series via time-frequency consistency. Adv. Neural. Inf. Process. Syst. 35, 3988–4003 (2022)
Cited by
Khan, F. U., & Mian, A. N. (2025) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2025.110753 · Cited by 5
Tanigaki, K., Maekawa, T., & Hara, T. (2026) · IEEE International Conference on Pervasive Computing and Communications · doi:10.1109/percom67906.2026.11524533 · Cited by 3
Developing Switching Mechanism to Address Challenging Subjects in Sleep Stage Scoring
Kawabata, R., Horie, K., Lalanda, P., & Kitagawa, H. (2025) · IEEE International Conference on Healthcare Informatics (ICHI) · doi:10.1109/ichi64645.2025.00064 · Cited by 1
Self-citationAI in Wearable Imaging Devices
Adebayo, P. O., Aremu, I. B., & Babatunde, A. J. (2025) · Advances in computational intelligence and robotics book series · doi:10.4018/979-8-3373-0903-3.ch006
Uncertainty-Driven Monitoring for ML-Based Autonomic Systems
Balditsyn, V., Lalanda, P., Vega, G., & Chollet, S. (2025) · IEEE International Conference on Autonomic Computing and Self-Organizing Systems (ACSOS) · doi:10.1109/acsos66086.2025.00021 · Cited by 1
“3: Model architecture combining MAE with a HAR Transformer [28]. their self-attention mechanisms effectively capture temporal dependencies and long-range relationships.”
Self-citation
Metadata
- Authors
- Ek, S., Presotto, R., Civitarese, G., Portet, F., Lalanda, P., & Bettini, C.
- Affiliations
- Centre National de la Recherche Scientifique (FR)
- University of Milan (IT)
- Corresponding author
- Ek, Sannara
- Venue
- CCF Transactions on Pervasive Computing and Interaction
- Published
- 2025-03-30
- Volume
- 7
- Issue
- 3
- Pages
- 324-341
- ISSN
- 2524-521X (ISSN-L)
- 2524-5228
- OpenAlex
- W4409016795
- Cited by
- 15
- References
- 57
- Citations by year
- 2026: 6 · 2025: 8
- FWCI
- 4.26
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Licence
- www.springernature.com/gp/researchers/text-and-data-mining (tdm)
- www.springernature.com/gp/researchers/text-and-data-mining (vor)
- Locations
- CCF Transactions on Pervasive Computing and Interaction (publishedVersion, public-domain)
- SSRN Electronic Journal (acceptedVersion)
- arXiv (Cornell University) (submittedVersion, public-domain)
- HAL (submittedVersion, cc-by)
- Archivio Istituzionale della Ricerca (Universita Degli Studi Di Milano) (submittedVersion, other-oa)
- arXiv (Cornell University)
- Funders
- Naval Group
- Nextgeneration
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Health Sciences › Medicine › Public Health, Environmental and Occupational Health › Physical Activity and Health
- Keywords
- human activity recognition, wearable sensors, self-supervised learning, contrastive learning, generative models, prediction model, masked autoencoder, SimCLR, limited labeled data, pre-trained models
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
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: Comparing self-supervised learning techniques for wearable human activity recognition. https://app.cochat.ai/papers/10.1007/s42486-024-00182-9
- The paper
- Ek, S., Presotto, R., Civitarese, G., Portet, F., Lalanda, P., & Bettini, C. (2025). Comparing self-supervised learning techniques for wearable human activity recognition. CCF Transactions on Pervasive Computing and Interaction. https://doi.org/10.1007/s42486-024-00182-9
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.