---
title: "SensorLLM — paper by Li et al. (2025) — summary and ideas"
description: "Large language models are rarely applied to wearable motion data. 65 references, cited by 6 papers here."
authors: ["Zechen Li", "Shohreh Deldari", "Linyao Chen", "Hao Xue", "Flora D. Salim"]
year: 2025
venue: "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing"
source: "https://aclanthology.org/2025.emnlp-main.19/"
doi: "10.18653/v1/2025.emnlp-main.19"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/li-2025-sensorllm-aligning-large-language-models/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/li-2025-sensorllm-aligning-large-language-models/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2025](https://md.research.elamur.ai/free/en/papers/2025/page/3/) / SensorLLM: Aligning Large Language Models with Motion Sensors for Human Activity Recognition

# SensorLLM: Aligning Large Language Models with Motion Sensors for Human Activity Recognition

**Authors:** Zechen Li, Shohreh Deldari, Linyao Chen, Hao Xue and Flora D. Salim · **Year:** 2025 · **Venue:** [Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing](https://md.research.elamur.ai/free/en/venues/emnlp/) ([EMNLP 2025](https://md.research.elamur.ai/free/en/venues/emnlp/2025/)) · **Publisher:** [Association for Computational Linguistics](https://md.research.elamur.ai/free/en/publishers/association-for-computational-linguistics/) · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2025.emnlp-main.19/) · [PDF](https://aclanthology.org/2025.emnlp-main.19.pdf) · [DOI 10.18653/v1/2025.emnlp-main.19](https://doi.org/10.18653/v1/2025.emnlp-main.19) · [OpenAlex](https://openalex.org/W4416037278)

**Citations:** 21 citations (OpenAlex, 2026-10-14) · [6](https://md.research.elamur.ai/free/en/papers/li-2025-sensorllm-aligning-large-language-models/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Large language models are rarely applied to wearable motion data. Raw sensor streams carry no semantic context, tokenizers handle numbers poorly, long multichannel sequences exceed context limits, and inter-channel structure is hard to preserve. SensorLLM addresses this with a two-stage design that leaves the LLM backbone and a pretrained time-series encoder frozen.

In the Sensor-Language Alignment stage, the encoder's embeddings pass through a small MLP into the LLM's input space. Training uses question-answer pairs built automatically from statistical trend analysis and randomized templates, which describe rising, falling and stable segments, so no human annotation is needed. Learned special tokens delimit where one sensor channel ends and the next begins. During the second stage, Task-Aware Tuning, human activity recognition (HAR) is carried out by a lightweight classifier placed on top of the LLM.

Across five HAR benchmarks, the method matches or beats state-of-the-art baselines. It handles inputs of varying length and channel count, and the learned alignment transfers between datasets. This points toward sensor–text foundation models that are interpretable and cheap to train.

## Key points
- Only a small share of parameters is trained: about 5.67% (535.9M) in alignment and 0.12% (10.5M) in task tuning.
- Evaluation datasets include UCI-HAR, PAMAP2, MHealth and USC-HAD, among five HAR benchmarks in total.
- Channel special tokens raise F1-macro on MHealth (89.6 → 90.2, 15 channels) and PAMAP2 (84.4 → 85.8, 27 channels). The gain grows with channel count.
- In cross-dataset tests, alignment was learned on USC-HAD and tuning done on UCI-HAR, or the reverse. Scores (about 91 F1 on UCI-HAR, about 61 on USC-HAD) were comparable to same-dataset training.
- A lighter SensorLLM-3b (Chronos-base + LLaMA3.2-3b) stays competitive. Code is released at github.com/zechenli03/SensorLLM.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

## Institutions

- [The University of Tokyo](https://md.research.elamur.ai/free/en/universities/the-university-of-tokyo/) — University, Tokyo, Japan · 90 works · 9,156 citations
- [UNSW Sydney](https://md.research.elamur.ai/free/en/universities/unsw-sydney/) — University, Sydney, Australia · 34 works · 1,509 citations

📚 Shelves (5): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [👁️ Computer Vision and Pattern Recognition](https://md.research.elamur.ai/free/en/shelves/computer-science/computer-vision-and-pattern-recognition/) · 2,869 works · [🤝 Social Sciences](https://md.research.elamur.ai/free/en/shelves/social-sciences/) · 1,772 works · [🔊 Signal Processing](https://md.research.elamur.ai/free/en/shelves/computer-science/signal-processing/) · 452 works · [🤝 Transportation](https://md.research.elamur.ai/free/en/shelves/social-sciences/transportation/) · 6 works

🧭 Topics (1): [🧠 Language models & pretraining](https://md.research.elamur.ai/free/en/topics/language-models/) · 3,320 works

## Cite

Cite as: Zechen Li, Shohreh Deldari, Linyao Chen, Hao Xue and Flora D. Salim. 2025. SensorLLM: Aligning Large Language Models with Motion Sensors for Human Activity Recognition. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{li2025sensorllm,
  title = {{SensorLLM: Aligning Large Language Models with Motion Sensors for Human Activity Recognition}},
  author = {Zechen Li, Shohreh Deldari, Linyao Chen, Hao Xue and Flora D. Salim},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  year = {2025},
  doi = {10.18653/v1/2025.emnlp-main.19},
  url = {https://doi.org/10.18653/v1/2025.emnlp-main.19}
}
```

- [← previous SciRIFF: A Resource to Enhance Language Model Instruction-Following over Scientific Literature](https://md.research.elamur.ai/free/en/papers/wadden-2025-sciriff-resource-enhance-language-model/)
- [next → Separating Tongue from Thought: Activation Patching Reveals Language-Agnostic Concept Representations in Transformers](https://md.research.elamur.ai/free/en/papers/dumas-2025-separating-tongue-thought-activation-patching/)
- [↑ back to papers of 2025](https://md.research.elamur.ai/free/en/papers/2025/page/3/)

## References

65 references, 11 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Alireza Abedin, Mahsa Ehsanpour, Qinfeng Shi, Hamid Rezatofighi, and Damith C. Ranasinghe. 2021. Attend and discriminate: Beyond the state-of-the-art for human activity recognition using wearable sensors. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., 5(1). [find](https://scholar.google.com/scholar?q=Alireza+Abedin%2C+Mahsa+Ehsanpour%2C+Qinfeng+Shi%2C+Hamid+Rezatofighi%2C+and+Damith+C%2E+Ranasinghe%2E+2021%2E+Attend+and+discriminate%3A+Beyond+the+state%2Dof%2Dthe%2Dart+for+human+activity+recognition+using+wearable+sens)
2. D. Anguita, Alessandro Ghio, L. Oneto, Xavier Parra, and Jorge Luis Reyes-Ortiz. 2013. A public domain dataset for human activity recognition using smartphones. In The European Symposium on Artificial Neural Networks. [find](https://scholar.google.com/scholar?q=D%2E+Anguita%2C+Alessandro+Ghio%2C+L%2E+Oneto%2C+Xavier+Parra%2C+and+Jorge+Luis+Reyes%2DOrtiz%2E+2013%2E+A+public+domain+dataset+for+human+activity+recognition+using+smartphones%2E+In+The+European+Symposium+on+Artificial)
3. Abdul Fatir Ansari, Lorenzo Stella, Caner Turkmen, Xiyuan Zhang, Pedro Mercado, Huibin Shen, Oleksandr Shchur, Syama Syndar Rangapuram, Sebastian Pineda Arango, Shubham Kapoor, Jasper Zschiegner, Danielle C. Maddix, Michael W. Mahoney, Kari Torkkola, Andrew Gordon Wilson, Michael Bohlke-Schneider, a… [find](https://scholar.google.com/scholar?q=Abdul+Fatir+Ansari%2C+Lorenzo+Stella%2C+Caner+Turkmen%2C+Xiyuan+Zhang%2C+Pedro+Mercado%2C+Huibin+Shen%2C+Oleksandr+Shchur%2C+Syama+Syndar+Rangapuram%2C+Sebastian+Pineda+Arango%2C+Shubham+Kapoor%2C+Jasper+Zschiegner%2C+Dani)
4. Satanjeev Banerjee and Alon Lavie. 2005. [METEOR: An automatic metric for MT evaluation with improved correlation with human judgments](https://md.research.elamur.ai/free/en/papers/banerjee-2005-meteor-automatic-metric-mt-evaluation/). In Proceedings of the ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization, pages 65–72, Ann Arbor, Michigan. Ass…
5. Nate Gruver, Marc Finzi, Shikai Qiu, and Andrew Gordon Wilson. 2023. Large language models are zero-shot time series forecasters. In Proceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23, Red Hook, NY, USA. Curran Associates Inc. [find](https://scholar.google.com/scholar?q=Nate+Gruver%2C+Marc+Finzi%2C+Shikai+Qiu%2C+and+Andrew+Gordon+Wilson%2E+2023%2E+Large+language+models+are+zero%2Dshot+time+series+forecasters%2E+In+Proceedings+of+the+37th+International+Conference+on+Neural+Informat)
6. Yu Guan and Thomas Plötz. 2017. Ensembles of deep lstm learners for activity recognition using wearables. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., 1(2). [find](https://scholar.google.com/scholar?q=Yu+Guan+and+Thomas+Pl%C3%B6tz%2E+2017%2E+Ensembles+of+deep+lstm+learners+for+activity+recognition+using+wearables%2E+Proc%2E+ACM+Interact%2E+Mob%2E+Wearable+Ubiquitous+Technol%2E%2C+1%282%29%2E)
7. Sojeong Ha and Seungjin Choi. 2016. Convolutional neural networks for human activity recognition using multiple accelerometer and gyroscope sensors. In 2016 International Joint Conference on Neural Networks (IJCNN), pages 381–388. [find](https://scholar.google.com/scholar?q=Sojeong+Ha+and+Seungjin+Choi%2E+2016%2E+Convolutional+neural+networks+for+human+activity+recognition+using+multiple+accelerometer+and+gyroscope+sensors%2E+In+2016+International+Joint+Conference+on+Neural+Ne)
8. Oresti Baños, Rafael García, Juan Antonio Holgado Terriza, Miguel Damas, Héctor Pomares, Ignacio Rojas, Alejandro Saez, and Claudia Villalonga. 2014. mhealthdroid: A novel framework for agile development of mobile health applications. In International Workshop on Ambient Assisted Living and Home Car… [find](https://scholar.google.com/scholar?q=Oresti+Ba%C3%B1os%2C+Rafael+Garc%C3%ADa%2C+Juan+Antonio+Holgado+Terriza%2C+Miguel+Damas%2C+H%C3%A9ctor+Pomares%2C+Ignacio+Rojas%2C+Alejandro+Saez%2C+and+Claudia+Villalonga%2E+2014%2E+mhealthdroid%3A+A+novel+framework+for+agile+developm)
9. Nils Y. Hammerla, Shane Halloran, and Thomas Plötz. 2016. Deep, convolutional, and recurrent models for human activity recognition using wearables. In Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence, IJCAI’16, page 1533–1540. AAAI Press. [find](https://scholar.google.com/scholar?q=Nils+Y%2E+Hammerla%2C+Shane+Halloran%2C+and+Thomas+Pl%C3%B6tz%2E+2016%2E+Deep%2C+convolutional%2C+and+recurrent+models+for+human+activity+recognition+using+wearables%2E+In+Proceedings+of+the+Twenty%2DFifth+International+Joi)
10. Antonio Bevilacqua, Kyle MacDonald, Aamina Ran-garej, Venessa Widjaya, Brian Caulfield, and Tahar Kechadi. 2019. Human Activity Recognition with Convolutional Neural Networks, page 541–552. Springer International Publishing. [find](https://scholar.google.com/scholar?q=Antonio+Bevilacqua%2C+Kyle+MacDonald%2C+Aamina+Ran%2Dgarej%2C+Venessa+Widjaya%2C+Brian+Caulfield%2C+and+Tahar+Kechadi%2E+2019%2E+Human+Activity+Recognition+with+Convolutional+Neural+Networks%2C+page+541%E2%80%93552%2E+Springer+I)
11. Xu Han, Zhengyan Zhang, Ning Ding, Yuxian Gu, Xiao Liu, Yuqi Huo, Jiezhong Qiu, Yuan Yao, Ao Zhang, Liang Zhang, Wentao Han, Minlie Huang, Qin Jin, Yanyan Lan, Yang Liu, Zhiyuan Liu, Zhiwu Lu, Xipeng Qiu, Ruihua Song, and 5 others. 2021. Pre-trained models: Past, present and future. Preprint, arXiv:… [find](https://scholar.google.com/scholar?q=Xu+Han%2C+Zhengyan+Zhang%2C+Ning+Ding%2C+Yuxian+Gu%2C+Xiao+Liu%2C+Yuqi+Huo%2C+Jiezhong+Qiu%2C+Yuan+Yao%2C+Ao+Zhang%2C+Liang+Zhang%2C+Wentao+Han%2C+Minlie+Huang%2C+Qin+Jin%2C+Yanyan+Lan%2C+Yang+Liu%2C+Zhiyuan+Liu%2C+Zhiwu+Lu%2C+Xipeng+)
12. Defu Cao, Furong Jia, Sercan O Arik, Tomas Pfister, Yixiang Zheng, Wen Ye, and Yan Liu. 2024. Tempo: Prompt-based generative pre-trained transformer for time series forecasting. Preprint, arXiv:2310.04948. — [arXiv:2310.04948](https://arxiv.org/abs/2310.04948) [find](https://scholar.google.com/scholar?q=Defu+Cao%2C+Furong+Jia%2C+Sercan+O+Arik%2C+Tomas+Pfister%2C+Yixiang+Zheng%2C+Wen+Ye%2C+and+Yan+Liu%2E+2024%2E+Tempo%3A+Prompt%2Dbased+generative+pre%2Dtrained+transformer+for+time+series+forecasting%2E+Preprint%2C+arXiv%3A2310%2E0)
13. Shing Chan, Hang Yuan, Catherine Tong, Aidan Ac-quah, Abram Schonfeldt, Jonathan Gershuny, and Aiden Doherty. 2024. Capture-24: A large dataset of wrist-worn activity tracker data collected in the wild for human activity recognition. Preprint, arXiv:2402.19229. — [arXiv:2402.19229](https://arxiv.org/abs/2402.19229) [find](https://scholar.google.com/scholar?q=Shing+Chan%2C+Hang+Yuan%2C+Catherine+Tong%2C+Aidan+Ac%2Dquah%2C+Abram+Schonfeldt%2C+Jonathan+Gershuny%2C+and+Aiden+Doherty%2E+2024%2E+Capture%2D24%3A+A+large+dataset+of+wrist%2Dworn+activity+tracker+data+collected+in+the+wil)
14. Ching Chang, Wei-Yao Wang, Wen-Chih Peng, and Tien-Fu Chen. 2024. Llm4ts: Aligning pre-trained llms as data-efficient time-series forecasters. Preprint, arXiv:2308.08469. — [arXiv:2308.08469](https://arxiv.org/abs/2308.08469) [find](https://scholar.google.com/scholar?q=Ching+Chang%2C+Wei%2DYao+Wang%2C+Wen%2DChih+Peng%2C+and+Tien%2DFu+Chen%2E+2024%2E+Llm4ts%3A+Aligning+pre%2Dtrained+llms+as+data%2Defficient+time%2Dseries+forecasters%2E+Preprint%2C+arXiv%3A2308%2E08469%2E+%E2%80%94+%5BarXiv%3A2308%2E08469%5D%28https%3A%2F%2F)
15. Shohreh Deldari, Dimitris Spathis, Mohammad Malekzadeh, Fahim Kawsar, Flora D. Salim, and Akhil Mathur. 2024. Crossl: Cross-modal self-supervised learning for time-series through latent masking. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining, WSDM ’24, page 152… [find](https://scholar.google.com/scholar?q=Shohreh+Deldari%2C+Dimitris+Spathis%2C+Mohammad+Malekzadeh%2C+Fahim+Kawsar%2C+Flora+D%2E+Salim%2C+and+Akhil+Mathur%2E+2024%2E+Crossl%3A+Cross%2Dmodal+self%2Dsupervised+learning+for+time%2Dseries+through+latent+masking%2E+In+Pr)
16. Tianyu Gao, Xingcheng Yao, and Danqi Chen. 2021. [SimCSE: Simple contrastive learning of sentence embeddings](https://md.research.elamur.ai/free/en/papers/gao-2021-simcse-simple-contrastive-learning-sentence/). In Proceedings of the 2021 Conference Harish Haresamudram, David V. Anderson, and Thomas Plötz. 2019. On the role of features in human activity recognition. In Proceedings of the 2019 ACM Int…
17. Harish Haresamudram, Apoorva Beedu, Mashfiqui Rabbi, Sankalita Saha, Irfan Essa, and Thomas Ploetz. 2025. Limitations in employing natural language supervision for sensor-based human activity recognition - and ways to overcome them. In Proceedings of the Thirty-Ninth AAAI Conference on Artificial In… [find](https://scholar.google.com/scholar?q=Harish+Haresamudram%2C+Apoorva+Beedu%2C+Mashfiqui+Rabbi%2C+Sankalita+Saha%2C+Irfan+Essa%2C+and+Thomas+Ploetz%2E+2025%2E+Limitations+in+employing+natural+language+supervision+for+sensor%2Dbased+human+activity+recognit)
18. Dan Hendrycks and Kevin Gimpel. 2016. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415. [find](https://scholar.google.com/scholar?q=Dan+Hendrycks+and+Kevin+Gimpel%2E+2016%2E+Gaussian+error+linear+units+%28gelus%29%2E+arXiv+preprint+arXiv%3A1606%2E08415%2E)
19. Sijie Ji, Xinzhe Zheng, and Chenshu Wu. 2024. Hargpt: Are llms zero-shot human activity recognizers? Preprint, arXiv:2403.02727. — [arXiv:2403.02727](https://arxiv.org/abs/2403.02727) [find](https://scholar.google.com/scholar?q=Sijie+Ji%2C+Xinzhe+Zheng%2C+and+Chenshu+Wu%2E+2024%2E+Hargpt%3A+Are+llms+zero%2Dshot+human+activity+recognizers%3F+Preprint%2C+arXiv%3A2403%2E02727%2E+%E2%80%94+%5BarXiv%3A2403%2E02727%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2403%2E02727%29)
20. Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, and Qingsong Wen. 2024a. Time-LLM: Time series forecasting by reprogramming large language models. In International Conference on Learning Representations (ICLR). [find](https://scholar.google.com/scholar?q=Ming+Jin%2C+Shiyu+Wang%2C+Lintao+Ma%2C+Zhixuan+Chu%2C+James+Y+Zhang%2C+Xiaoming+Shi%2C+Pin%2DYu+Chen%2C+Yuxuan+Liang%2C+Yuan%2DFang+Li%2C+Shirui+Pan%2C+and+Qingsong+Wen%2E+2024a%2E+Time%2DLLM%3A+Time+series+forecasting+by+reprogramm)
21. Che Liu, Zhongwei Wan, Sibo Cheng, Mi Zhang, and Rossella Arcucci. 2024a. Etp: Learning transferable ecg representations via ecg-text pre-training. In ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 8230–8234. [find](https://scholar.google.com/scholar?q=Che+Liu%2C+Zhongwei+Wan%2C+Sibo+Cheng%2C+Mi+Zhang%2C+and+Rossella+Arcucci%2E+2024a%2E+Etp%3A+Learning+transferable+ecg+representations+via+ecg%2Dtext+pre%2Dtraining%2E+In+ICASSP+2024+%2D+2024+IEEE+International+Conference+)
22. Ming Jin, Qingsong Wen, Yuxuan Liang, Chaoli Zhang, Siqiao Xue, Xue Wang, James Zhang, Yi Wang, Haifeng Chen, Xiaoli Li, Shirui Pan, Vincent S. Tseng, Yu Zheng, Lei Chen, and Hui Xiong. 2023. Large models for time series and spatiotemporal data: A survey and outlook. Preprint, arXiv:2310.10196. — [arXiv:2310.10196](https://arxiv.org/abs/2310.10196) [find](https://scholar.google.com/scholar?q=Ming+Jin%2C+Qingsong+Wen%2C+Yuxuan+Liang%2C+Chaoli+Zhang%2C+Siqiao+Xue%2C+Xue+Wang%2C+James+Zhang%2C+Yi+Wang%2C+Haifeng+Chen%2C+Xiaoli+Li%2C+Shirui+Pan%2C+Vincent+S%2E+Tseng%2C+Yu+Zheng%2C+Lei+Chen%2C+and+Hui+Xiong%2E+2023%2E+Large+mo)
23. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023a. [Visual instruction tuning](https://md.research.elamur.ai/free/en/papers/liu-2023-visual-instruction-tuning/). In Advances in Neural Information Processing Systems, volume 36, pages 34892–34916. Curran Associates, Inc.
24. Ming Jin, Yifan Zhang, Wei Chen, Kexin Zhang, Yuxuan Liang, Bin Yang, Jindong Wang, Shirui Pan, and Qingsong Wen. 2024b. Position paper: What can large language models tell us about time series analysis. Preprint, arXiv:2402.02713. [find](https://scholar.google.com/scholar?q=Ming+Jin%2C+Yifan+Zhang%2C+Wei+Chen%2C+Kexin+Zhang%2C+Yuxuan+Liang%2C+Bin+Yang%2C+Jindong+Wang%2C+Shirui+Pan%2C+and+Qingsong+Wen%2E+2024b%2E+Position+paper%3A+What+can+large+language+models+tell+us+about+time+series+analys)
25. Panagiotis Kasnesis, Charalampos Z. Patrikakis, and Iakovos S. Venieris. 2019. Perceptionnet: A deep convolutional neural network for late sensor fusion. In Intelligent Systems and Applications, pages 101– 119, Cham. Springer International Publishing. [find](https://scholar.google.com/scholar?q=Panagiotis+Kasnesis%2C+Charalampos+Z%2E+Patrikakis%2C+and+Iakovos+S%2E+Venieris%2E+2019%2E+Perceptionnet%3A+A+deep+convolutional+neural+network+for+late+sensor+fusion%2E+In+Intelligent+Systems+and+Applications%2C+pages)
26. Yubin Kim, Xuhai Xu, Daniel McDuff, Cynthia Breazeal, and Hae Won Park. 2024. Health-llm: Large language models for health prediction via wearable sensor data. In Proceedings of the fifth Conference on Health, Inference, and Learning, volume 248 of Proceedings of Machine Learning Research, pages 522… [find](https://scholar.google.com/scholar?q=Yubin+Kim%2C+Xuhai+Xu%2C+Daniel+McDuff%2C+Cynthia+Breazeal%2C+and+Hae+Won+Park%2E+2024%2E+Health%2Dllm%3A+Large+language+models+for+health+prediction+via+wearable+sensor+data%2E+In+Proceedings+of+the+fifth+Conference+o)
27. Jennifer R. Kwapisz, Gary M. Weiss, and Samuel A. Moore. 2011. Activity recognition using cell phone accelerometers. SIGKDD Explor. Newsl., 12(2):74–82. [find](https://scholar.google.com/scholar?q=Jennifer+R%2E+Kwapisz%2C+Gary+M%2E+Weiss%2C+and+Samuel+A%2E+Moore%2E+2011%2E+Activity+recognition+using+cell+phone+accelerometers%2E+SIGKDD+Explor%2E+Newsl%2E%2C+12%282%29%3A74%E2%80%9382%2E)
28. Xin Liu, Daniel McDuff, Geza Kovacs, Isaac Galatzer-Levy, Jacob Sunshine, Jiening Zhan, Ming-Zher Poh, Shun Liao, Paolo Di Achille, and Shwetak Patel. 2023b. Large language models are few-shot health learners. Preprint, arXiv:2305.15525. — [arXiv:2305.15525](https://arxiv.org/abs/2305.15525) [find](https://scholar.google.com/scholar?q=Xin+Liu%2C+Daniel+McDuff%2C+Geza+Kovacs%2C+Isaac+Galatzer%2DLevy%2C+Jacob+Sunshine%2C+Jiening+Zhan%2C+Ming%2DZher+Poh%2C+Shun+Liao%2C+Paolo+Di+Achille%2C+and+Shwetak+Patel%2E+2023b%2E+Large+language+models+are+few%2Dshot+health+)
29. Xu Liu, Junfeng Hu, Yuan Li, Shizhe Diao, Yuxuan Liang, Bryan Hooi, and Roger Zimmermann. 2024b. Unitime: A language-empowered unified model for cross-domain time series forecasting. In Proceedings of the ACM Web Conference 2024, WWW ’24, page 4095–4106, New York, NY, USA. Association for Computing… [find](https://scholar.google.com/scholar?q=Xu+Liu%2C+Junfeng+Hu%2C+Yuan+Li%2C+Shizhe+Diao%2C+Yuxuan+Liang%2C+Bryan+Hooi%2C+and+Roger+Zimmermann%2E+2024b%2E+Unitime%3A+A+language%2Dempowered+unified+model+for+cross%2Ddomain+time+series+forecasting%2E+In+Proceedings+of)
30. Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. 2024c. itransformer: Inverted transformers are effective for time series forecasting. International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Yong+Liu%2C+Tengge+Hu%2C+Haoran+Zhang%2C+Haixu+Wu%2C+Shiyu+Wang%2C+Lintao+Ma%2C+and+Mingsheng+Long%2E+2024c%2E+itransformer%3A+Inverted+transformers+are+effective+for+time+series+forecasting%2E+International+Conference+o)
31. Yong Liu, Haixu Wu, Jianmin Wang, and Mingsheng Long. 2022. Non-stationary transformers: exploring the stationarity in time series forecasting. In Proceedings of the 36th International Conference on Neural Information Processing Systems, NIPS ’22, Red Hook, NY, USA. Curran Associates Inc. [find](https://scholar.google.com/scholar?q=Yong+Liu%2C+Haixu+Wu%2C+Jianmin+Wang%2C+and+Mingsheng+Long%2E+2022%2E+Non%2Dstationary+transformers%3A+exploring+the+stationarity+in+time+series+forecasting%2E+In+Proceedings+of+the+36th+International+Conference+on+N)
32. Zikang Leng, Hyeokhyen Kwon, and Thomas Ploetz. 2023. Generating virtual on-body accelerometer data from virtual textual descriptions for human activity recognition. In Proceedings of the 2023 ACM International Symposium on Wearable Computers, ISWC ’23. [find](https://scholar.google.com/scholar?q=Zikang+Leng%2C+Hyeokhyen+Kwon%2C+and+Thomas+Ploetz%2E+2023%2E+Generating+virtual+on%2Dbody+accelerometer+data+from+virtual+textual+descriptions+for+human+activity+recognition%2E+In+Proceedings+of+the+2023+ACM+Int)
33. Seungwhan Moon, Andrea Madotto, Zhaojiang Lin, Aparajita Saraf, Amy Bearman, and Babak Damavandi. 2023. IMU2CLIP: Language-grounded motion sensor translation with multimodal contrastive learning. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 13246– 13253, Singapore.… [find](https://scholar.google.com/scholar?q=Seungwhan+Moon%2C+Andrea+Madotto%2C+Zhaojiang+Lin%2C+Aparajita+Saraf%2C+Amy+Bearman%2C+and+Babak+Damavandi%2E+2023%2E+IMU2CLIP%3A+Language%2Dgrounded+motion+sensor+translation+with+multimodal+contrastive+learning%2E+In+F)
34. Hong Li, Gregory D. Abowd, and Thomas Plötz. 2018. On specialized window lengths and detector based human activity recognition. In Proceedings of the 2018 ACM International Symposium on Wearable Computers, ISWC ’18, page 68–71, New York, NY, USA. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Hong+Li%2C+Gregory+D%2E+Abowd%2C+and+Thomas+Pl%C3%B6tz%2E+2018%2E+On+specialized+window+lengths+and+detector+based+human+activity+recognition%2E+In+Proceedings+of+the+2018+ACM+International+Symposium+on+Wearable+Compu)
35. Vishvak S. Murahari and Thomas Plötz. 2018. On attention models for human activity recognition. In Proceedings of the 2018 ACM International Symposium on Wearable Computers, ISWC ’18, page 100–103, New York, NY, USA. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Vishvak+S%2E+Murahari+and+Thomas+Pl%C3%B6tz%2E+2018%2E+On+attention+models+for+human+activity+recognition%2E+In+Proceedings+of+the+2018+ACM+International+Symposium+on+Wearable+Computers%2C+ISWC+%E2%80%9918%2C+page+100%E2%80%93103%2C+Ne)
36. Zechen Li, Baiyu Chen, Hao Xue, and Flora D. Salim. 2025. Zara: Zero-shot motion time-series analysis via knowledge and retrieval driven llm agents. Preprint, arXiv:2508.04038. — [arXiv:2508.04038](https://arxiv.org/abs/2508.04038) [find](https://scholar.google.com/scholar?q=Zechen+Li%2C+Baiyu+Chen%2C+Hao+Xue%2C+and+Flora+D%2E+Salim%2E+2025%2E+Zara%3A+Zero%2Dshot+motion+time%2Dseries+analysis+via+knowledge+and+retrieval+driven+llm+agents%2E+Preprint%2C+arXiv%3A2508%2E04038%2E+%E2%80%94+%5BarXiv%3A2508%2E04038%5D%28ht)
37. Chin-Yew Lin. 2004. [ROUGE: A package for automatic evaluation of summaries](https://md.research.elamur.ai/free/en/papers/lin-2004-rouge-package-automatic-evaluation-summaries/). In Text Summarization Branches Out, pages 74–81, Barcelona, Spain. Association for Computational Linguistics.
38. Shikai Qiu Nate Gruver, Marc Finzi and Andrew Gordon Wilson. 2023. Large Language Models Are Zero Shot Time Series Forecasters. In Advances in Neural Information Processing Systems. [find](https://scholar.google.com/scholar?q=Shikai+Qiu+Nate+Gruver%2C+Marc+Finzi+and+Andrew+Gordon+Wilson%2E+2023%2E+Large+Language+Models+Are+Zero+Shot+Time+Series+Forecasters%2E+In+Advances+in+Neural+Information+Processing+Systems%2E)
39. Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2023. A time series is worth 64 words: Long-term forecasting with transformers. In International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Yuqi+Nie%2C+Nam+H%2E+Nguyen%2C+Phanwadee+Sinthong%2C+and+Jayant+Kalagnanam%2E+2023%2E+A+time+series+is+worth+64+words%3A+Long%2Dterm+forecasting+with+transformers%2E+In+International+Conference+on+Learning+Representati)
40. OpenAI. 2024. [Gpt-4 technical report](https://md.research.elamur.ai/free/en/papers/openai-2023-gpt-4-technical-report/). arXiv:2303.08774.
41. Preprint, Francisco Javier Ordóñez and Daniel Roggen. 2016. Deep convolutional and lstm recurrent neural networks for multimodal wearable activity recognition. Sensors, 16(1). [find](https://scholar.google.com/scholar?q=Preprint%2C+Francisco+Javier+Ord%C3%B3%C3%B1ez+and+Daniel+Roggen%2E+2016%2E+Deep+convolutional+and+lstm+recurrent+neural+networks+for+multimodal+wearable+activity+recognition%2E+Sensors%2C+16%281%29%2E)
42. Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. [Bleu: a method for automatic evaluation of machine translation](https://md.research.elamur.ai/free/en/papers/papineni-2002-bleu-method-automatic-evaluation-machine/). In Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, ACL ’02, page 311–318, USA. Association for Computational Linguistics.
43. Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. OpenAI. [find](https://scholar.google.com/scholar?q=Alec+Radford%2C+Jeff+Wu%2C+Rewon+Child%2C+David+Luan%2C+Dario+Amodei%2C+and+Ilya+Sutskever%2E+2019%2E+Language+models+are+unsupervised+multitask+learners%2E+OpenAI%2E)
44. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. 2020. [Exploring the limits of transfer learning with a unified text-to-text transformer](https://md.research.elamur.ai/free/en/papers/raffel-2019-exploring-limits-transfer-learning-unified/). Journal of Machine Learning Research, 21(140):1–67.
45. Nils Reimers and Iryna Gurevych. 2019. [Sentence-bert: Sentence embeddings using siamese bert-networks](https://md.research.elamur.ai/free/en/papers/reimers-2019-sentence-bert-sentence-embeddings-siamese/). In Conference on Empirical Methods in Natural Language Processing.
46. Attila Reiss and Didier Stricker. 2012. Introducing a new benchmarked dataset for activity monitoring. In 2012 16th International Symposium on Wearable Computers, pages 108–109. [find](https://scholar.google.com/scholar?q=Attila+Reiss+and+Didier+Stricker%2E+2012%2E+Introducing+a+new+benchmarked+dataset+for+activity+monitoring%2E+In+2012+16th+International+Symposium+on+Wearable+Computers%2C+pages+108%E2%80%93109%2E)
47. Dimitris Spathis and Fahim Kawsar. 2023. The first step is the hardest: Pitfalls of representing and tokenizing temporal data for large language models. Preprint, arXiv:2309.06236. — [arXiv:2309.06236](https://arxiv.org/abs/2309.06236) [find](https://scholar.google.com/scholar?q=Dimitris+Spathis+and+Fahim+Kawsar%2E+2023%2E+The+first+step+is+the+hardest%3A+Pitfalls+of+representing+and+tokenizing+temporal+data+for+large+language+models%2E+Preprint%2C+arXiv%3A2309%2E06236%2E+%E2%80%94+%5BarXiv%3A2309%2E06236)
48. Chenxi Sun, Hongyan Li, Yaliang Li, and Shenda Hong. 2024. Test: Text prototype aligned embedding to activate llm’s ability for time series. International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Chenxi+Sun%2C+Hongyan+Li%2C+Yaliang+Li%2C+and+Shenda+Hong%2E+2024%2E+Test%3A+Text+prototype+aligned+embedding+to+activate+llm%E2%80%99s+ability+for+time+series%2E+International+Conference+on+Learning+Representations%2E)
49. Mingtian Tan, Mike A. Merrill, Vinayak Gupta, Tim Althoff, and Thomas Hartvigsen. 2024. Are language models actually useful for time series forecasting? Preprint, arXiv:2406.16964. — [arXiv:2406.16964](https://arxiv.org/abs/2406.16964) [find](https://scholar.google.com/scholar?q=Mingtian+Tan%2C+Mike+A%2E+Merrill%2C+Vinayak+Gupta%2C+Tim+Althoff%2C+and+Thomas+Hartvigsen%2E+2024%2E+Are+language+models+actually+useful+for+time+series+forecasting%3F+Preprint%2C+arXiv%3A2406%2E16964%2E+%E2%80%94+%5BarXiv%3A2406%2E16964)
50. Haixu Wu, Tengge Hu, Yong Liu, Hang Zhou, Jianmin Wang, and Mingsheng Long. 2023. Timesnet: Temporal 2d-variation modeling for general time series analysis. In International Conference on Learning Representations. [find](https://scholar.google.com/scholar?q=Haixu+Wu%2C+Tengge+Hu%2C+Yong+Liu%2C+Hang+Zhou%2C+Jianmin+Wang%2C+and+Mingsheng+Long%2E+2023%2E+Timesnet%3A+Temporal+2d%2Dvariation+modeling+for+general+time+series+analysis%2E+In+International+Conference+on+Learning+Rep)
51. Shengqiong Wu, Hao Fei, Leigang Qu, Wei Ji, and Tat-Seng Chua. 2024. Next-gpt: any-to-any multimodal llm. In Proceedings of the 41st International Conference on Machine Learning, ICML’24. JMLR.org. [find](https://scholar.google.com/scholar?q=Shengqiong+Wu%2C+Hao+Fei%2C+Leigang+Qu%2C+Wei+Ji%2C+and+Tat%2DSeng+Chua%2E+2024%2E+Next%2Dgpt%3A+any%2Dto%2Dany+multimodal+llm%2E+In+Proceedings+of+the+41st+International+Conference+on+Machine+Learning%2C+ICML%E2%80%9924%2E+JMLR%2Eorg%2E)
52. Kang Xia, Wenzhong Li, Shiwei Gan, and Sanglu Lu. 2024. Ts2act: Few-shot human activity sensing with cross-modal co-learning. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., 7(4). [find](https://scholar.google.com/scholar?q=Kang+Xia%2C+Wenzhong+Li%2C+Shiwei+Gan%2C+and+Sanglu+Lu%2E+2024%2E+Ts2act%3A+Few%2Dshot+human+activity+sensing+with+cross%2Dmodal+co%2Dlearning%2E+Proc%2E+ACM+Interact%2E+Mob%2E+Wearable+Ubiquitous+Technol%2E%2C+7%284%29%2E)
53. Qingxin Xia, Takuya Maekawa, and Takahiro Hara. 2023. Unsupervised human activity recognition through two-stage prompting with chatgpt. Preprint, arXiv:2306.02140. — [arXiv:2306.02140](https://arxiv.org/abs/2306.02140) [find](https://scholar.google.com/scholar?q=Qingxin+Xia%2C+Takuya+Maekawa%2C+and+Takahiro+Hara%2E+2023%2E+Unsupervised+human+activity+recognition+through+two%2Dstage+prompting+with+chatgpt%2E+Preprint%2C+arXiv%3A2306%2E02140%2E+%E2%80%94+%5BarXiv%3A2306%2E02140%5D%28https%3A%2F%2Farxiv%2Eo)
54. Cheng Xu, Duo Chai, Jie He, Xiaotong Zhang, and Shihong Duan. 2019. Innohar: A deep neural network for complex human activity recognition. IEEE Access, 7:9893–9902. [find](https://scholar.google.com/scholar?q=Cheng+Xu%2C+Duo+Chai%2C+Jie+He%2C+Xiaotong+Zhang%2C+and+Shihong+Duan%2E+2019%2E+Innohar%3A+A+deep+neural+network+for+complex+human+activity+recognition%2E+IEEE+Access%2C+7%3A9893%E2%80%939902%2E)
55. Hao Xue and Flora D. Salim. 2024. Promptcast: A new prompt-based learning paradigm for time series forecasting. IEEE Transactions on Knowledge and Data Engineering, 36(11):6851–6864. [find](https://scholar.google.com/scholar?q=Hao+Xue+and+Flora+D%2E+Salim%2E+2024%2E+Promptcast%3A+A+new+prompt%2Dbased+learning+paradigm+for+time+series+forecasting%2E+IEEE+Transactions+on+Knowledge+and+Data+Engineering%2C+36%2811%29%3A6851%E2%80%936864%2E)
56. Gilad Yehudai, Haim Kaplan, Asma Ghandeharioun, Mor Geva, and Amir Globerson. 2024. When can transformers count to n? Preprint, arXiv:2407.15160. — [arXiv:2407.15160](https://arxiv.org/abs/2407.15160) [find](https://scholar.google.com/scholar?q=Gilad+Yehudai%2C+Haim+Kaplan%2C+Asma+Ghandeharioun%2C+Mor+Geva%2C+and+Amir+Globerson%2E+2024%2E+When+can+transformers+count+to+n%3F+Preprint%2C+arXiv%3A2407%2E15160%2E+%E2%80%94+%5BarXiv%3A2407%2E15160%5D%28https%3A%2F%2Farxiv%2Eorg%2Fabs%2F2407%2E15160%29)
57. Shukang Yin, Chaoyou Fu, Sirui Zhao, Ke Li, Xing Sun, Tong Xu, and Enhong Chen. 2024. [A survey on multimodal large language models](https://md.research.elamur.ai/free/en/papers/yin-2023-survey-multimodal-large-language-models/). National Science Review, 11(12).
58. Hyungjun Yoon, Biniyam Aschalew Tolera, Taesik Gong, Kimin Lee, and Sung-Ju Lee. 2024. By my eyes: Grounding multimodal large language models with sensor data via visual prompting. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 2219–2241, Miami, Flor… [find](https://scholar.google.com/scholar?q=Hyungjun+Yoon%2C+Biniyam+Aschalew+Tolera%2C+Taesik+Gong%2C+Kimin+Lee%2C+and+Sung%2DJu+Lee%2E+2024%2E+By+my+eyes%3A+Grounding+multimodal+large+language+models+with+sensor+data+via+visual+prompting%2E+In+Proceedings+of+t)
59. Yuta Yuki, Junto Nozaki, Kei Hiroi, Katsuhiko Kaji, and Nobuo Kawaguchi. 2018. Activity recognition using dual-convlstm extracting local and global features for shl recognition challenge. In Proceedings of the 2018 ACM International Joint Conference and 2018 International Symposium on Pervasive and… [find](https://scholar.google.com/scholar?q=Yuta+Yuki%2C+Junto+Nozaki%2C+Kei+Hiroi%2C+Katsuhiko+Kaji%2C+and+Nobuo+Kawaguchi%2E+2018%2E+Activity+recognition+using+dual%2Dconvlstm+extracting+local+and+global+features+for+shl+recognition+challenge%2E+In+Proceedin)
60. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, and Guillaume Lample. 2023. [Llama: Open and efficient foundation language models](https://md.research.elamur.ai/free/en/papers/touvron-2023-llama-open-efficient-foundation-language/). Preprint…
61. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. 2017. [Attention is all you need](https://md.research.elamur.ai/free/en/papers/vaswani-2017-attention-all-need/). In Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc.
62. Mi Zhang and Alexander A. Sawchuk. 2012. Usc-had: a daily activity dataset for ubiquitous activity recognition using wearable sensors. In Proceedings of the 2012 ACM Conference on Ubiquitous Computing, UbiComp ’12, page 1036–1043, New York, NY, USA. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Mi+Zhang+and+Alexander+A%2E+Sawchuk%2E+2012%2E+Usc%2Dhad%3A+a+daily+activity+dataset+for+ubiquitous+activity+recognition+using+wearable+sensors%2E+In+Proceedings+of+the+2012+ACM+Conference+on+Ubiquitous+Computing)
63. Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. 2021. Informer: Beyond efficient transformer for long sequence time-series forecasting. In The Thirty-Fifth AAAI Conference on Artificial Intelligence, AAAI 2021, Virtual Conference. AAAI Press. [find](https://scholar.google.com/scholar?q=Haoyi+Zhou%2C+Shanghang+Zhang%2C+Jieqi+Peng%2C+Shuai+Zhang%2C+Jianxin+Li%2C+Hui+Xiong%2C+and+Wancai+Zhang%2E+2021%2E+Informer%3A+Beyond+efficient+transformer+for+long+sequence+time%2Dseries+forecasting%2E+In+The+Thirty%2DFif)
64. Tian Zhou, Peisong Niu, Xue Wang, Liang Sun, and Rong Jin. 2023a. One fits all: power general time series analysis by pretrained lm. In Proceedings of the 37th International Conference on Neural Information Processing Systems, NIPS ’23, Red Hook, NY, USA. Curran Associates Inc. [find](https://scholar.google.com/scholar?q=Tian+Zhou%2C+Peisong+Niu%2C+Xue+Wang%2C+Liang+Sun%2C+and+Rong+Jin%2E+2023a%2E+One+fits+all%3A+power+general+time+series+analysis+by+pretrained+lm%2E+In+Proceedings+of+the+37th+International+Conference+on+Neural+Infor)
65. Yunjiao Zhou, Jianfei Yang, Han Zou, and Lihua Xie. 2023b. Tent: Connect language models with iot sensors for zero-shot activity recognition. Preprint, arXiv:2311.08245. — [arXiv:2311.08245](https://arxiv.org/abs/2311.08245) [find](https://scholar.google.com/scholar?q=Yunjiao+Zhou%2C+Jianfei+Yang%2C+Han+Zou%2C+and+Lihua+Xie%2E+2023b%2E+Tent%3A+Connect+language+models+with+iot+sensors+for+zero%2Dshot+activity+recognition%2E+Preprint%2C+arXiv%3A2311%2E08245%2E+%E2%80%94+%5BarXiv%3A2311%2E08245%5D%28https%3A%2F%2Fa)

*Updated 31 October 2026*