Harish Haresamudram

I am a Postdoctoral Research Associate at the Grainger College of Engineering, University of Illinois Urbana-Champaign, working under the mentorship of Prof. Jim Rehg. I obtained my PhD from Georgia Institute of Technology, where I was advised by Prof. Thomas Ploetz and Prof. Irfan Essa. My dissertation involved the design and development of representation learning approaches for wearable sensors, e.g., accelerometers and gyroscopes, for tasks such as Human Activity Recognition (HAR) and behavior analysis. I have been supported by funding from the National Science Foundation (through the AI-CARING Institute), Optum AI, and Google.

Research

My research broadly involves learning representations for time-series data, with a focus on developing techniques that require minimal supervision. I develop self-supervised and multi-modal learning algorithms for data from wearable sensors, e.g., accelerometers and PPG. Subsequently, I use such representations to analyse health and well-being as well as human behavior.

News

Mar 06, 2026	Our new paper: How Well Do Multimodal Models Reason on ECG Signals? is now available on Arxiv! We introduce a reproducible framework for evaluating reasoning capablities of Multimodal LLMs on ECG signals.
Nov 19, 2025	Our paper: Models Got Talent: Identifying High Performing Wearable Human Activity Recognition Models Without Training is now on Arxiv! We evaluate whether activity recognition performance can be effectively predicted without doing any training, leading to lightweight NAS.
Oct 15, 2025	Honored to be the runner-up for the Gaetano Boriello Outstanding Award at Ubicomp 2025.
Oct 12, 2025	Organized the GenAI4HS workshop at Ubicomp 2025! Great to see a packed session and interesting discussions!
Oct 01, 2025	I joined as a Post Doctoral Research Associate in Prof. Jim Rehg’s lab at UIUC! Excited for what is to come!

Selected publications

How Well Do Multimodal Models Reason on ECG Signals?

Maxwell A Xu, Harish Haresamudram, Catherine W Liu, and 8 more authors

arXiv preprint arXiv:2603.00312, 2026

Bib HTML PDF

@article{xu2026well,
  title = {How Well Do Multimodal Models Reason on ECG Signals?},
  author = {Xu, Maxwell A and Haresamudram, Harish and Liu, Catherine W and Langer, Patrick and Pradeepkumar, Jathurshan and Mao, Wanting and Ferns, Sunita J and Verma, Aradhana and Sun, Jimeng and Schmiedmayer, Paul and others},
  journal = {arXiv preprint arXiv:2603.00312},
  year = {2026},
  url = {https://arxiv.org/abs/2603.00312},
}

Models Got Talent: Identifying High Performing Wearable Human Activity Recognition Models Without Training

Richard Goldman, Varun Komperla, Thomas Ploetz, and 1 more author

arXiv preprint arXiv:2511.06157, 2025

Bib HTML PDF

@article{goldman2025models,
  title = {Models Got Talent: Identifying High Performing Wearable Human Activity Recognition Models Without Training},
  author = {Goldman, Richard and Komperla, Varun and Ploetz, Thomas and Haresamudram, Harish},
  journal = {arXiv preprint arXiv:2511.06157},
  year = {2025},
  url = {https://arxiv.org/abs/2511.06157},
}

Past, present, and future of sensor-based human activity recognition using wearables: A surveying tutorial on a still challenging task

Harish Haresamudram, Chi Ian Tang, Sungho Suh, and 2 more authors

Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2025

Bib HTML PDF

@article{haresamudram2025past,
  title = {Past, present, and future of sensor-based human activity recognition using wearables: A surveying tutorial on a still challenging task},
  author = {Haresamudram, Harish and Tang, Chi Ian and Suh, Sungho and Lukowicz, Paul and Ploetz, Thomas},
  journal = {Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies},
  volume = {9},
  number = {2},
  pages = {1--44},
  year = {2025},
  publisher = {ACM New York, NY, USA},
  url = {https://dl.acm.org/doi/pdf/10.1145/3729467},
}

Limitations in Employing Natural Language Supervision for Sensor-Based Human Activity Recognition-And Ways to Overcome Them

Harish Haresamudram, Apoorva Beedu, Mashfiqui Rabbi, and 3 more authors

In Proceedings of the AAAI Conference on Artificial Intelligence, 2025

Bib HTML PDF

@inproceedings{haresamudram2025limitations,
  title = {Limitations in Employing Natural Language Supervision for Sensor-Based Human Activity Recognition-And Ways to Overcome Them},
  author = {Haresamudram, Harish and Beedu, Apoorva and Rabbi, Mashfiqui and Saha, Sankalita and Essa, Irfan and Ploetz, Thomas},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume = {39},
  number = {1},
  pages = {273--281},
  year = {2025},
  url = {https://ojs.aaai.org/index.php/AAAI/article/view/32004},
}

Layout-agnostic human activity recognition in smart homes through textual descriptions of sensor triggers (TDOST)

Megha Thukral, Sourish Gunesh Dhekane, Shruthi K Hiremath, and 2 more authors

Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 2025

Bib HTML PDF

@article{thukral2025layout,
  title = {Layout-agnostic human activity recognition in smart homes through textual descriptions of sensor triggers (TDOST)},
  author = {Thukral, Megha and Dhekane, Sourish Gunesh and Hiremath, Shruthi K and Haresamudram, Harish and Ploetz, Thomas},
  journal = {Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies},
  volume = {9},
  number = {1},
  pages = {1--38},
  year = {2025},
  publisher = {ACM New York, NY, USA},
  url = {https://dl.acm.org/doi/pdf/10.1145/3712278},
}

Towards Learning Discrete Representations via Self-Supervision for Wearables-Based Human Activity Recognition

Harish Haresamudram, Irfan Essa, and Thomas Ploetz

Sensors, 2024

Bib

@article{haresamudram2024towards,
  title = {Towards Learning Discrete Representations via Self-Supervision for Wearables-Based Human Activity Recognition},
  author = {Haresamudram, Harish and Essa, Irfan and Ploetz, Thomas},
  journal = {Sensors},
  volume = {24},
  number = {4},
  pages = {1238},
  year = {2024},
  publisher = {Multidisciplinary Digital Publishing Institute},
}