Chong Tang

I am Chong Tang, an Assistant Professor (UK Lecturer) in Artificial Intelligence at the University of Bristol, in the School of Engineering Mathematics and Technology.

I build models that reason reliably across modalities and run efficiently in the real world. My research spans four connected themes:

Multimodal reasoning — reasoning across vision, language, and other modalities.
Representation learning — learning world models that capture the structure and dynamics of the world.
Agentic AI — models that plan, act, and adapt.
Efficient AI — making models fast and resource-aware enough to deploy.

I have published 30+ peer-reviewed papers at top venues in AI and machine learning. Previously, I was a postdoctoral researcher at UCL and the University of Southampton (with Dr. Jagmohan Chauhan and Dr. Alex Weddell), and I completed my PhD at UCL, advised by Prof. Kevin Chetty and Prof. Simon Julier.

I am recruiting PhD students. If you are excited about multimodal reasoning, representation learning, agentic AI, or efficient AI, I would love to hear from you. Email me at chong.tang@bristol.ac.uk with a short note about your interests and your CV — and browse my projects and publications.

Academic Service

Program Committee: ICMR 2026; EuroSys 2026 (Shadow PC); MobiUK 2024.

Journal Reviewer: IEEE Transactions on Mobile Computing; IEEE Sensors Journal; IEEE Internet of Things Journal; IEEE Journal of Biomedical and Health Informatics.

Conference Reviewer: NeurIPS; ICLR; ICML; ACM Multimedia; IEEE PerCom; UbiComp.

news

Jun 01, 2026	Joined the University of Bristol as an Assistant Professor (UK Lecturer) in AI, in the School of Engineering Mathematics and Technology.
Mar 01, 2026	Awarded AIRR funding for a project on efficient multimodal reasoning.
Jan 23, 2026	Our paper SURGE (surprise-guided token reduction for efficient video understanding with VLMs) was accepted at ICLR 2026.
Sep 26, 2025	PhySwin, our efficient, physically-grounded foundation model for multispectral Earth observation, was accepted at NeurIPS 2025.
Jun 01, 2025	Received an NVIDIA Academic Grant for a cloud-edge collaborative data generation framework for VLA learning.

selected publications

ICLR

SURGE: Surprise-Guided Token Reduction for Efficient Video Understanding with VLMs

Chong Tang, S. Ek, D. Koch, R. D. Mullins, A. S. Weddell, and J. Chauhan

In International Conference on Learning Representations (ICLR), 2026

Bib

@inproceedings{tang2026surge,
  title = {SURGE: Surprise-Guided Token Reduction for Efficient Video Understanding with VLMs},
  author = {Tang, Chong and Ek, S. and Koch, D. and Mullins, R. D. and Weddell, A. S. and Chauhan, J.},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2026},
}

NeurIPS

PhySwin: An Efficient and Physically-Informed Foundation Model for Multispectral Earth Observation

Chong Tang, J. Powell, D. Koch, R. D. Mullins, A. S. Weddell, and J. Chauhan

In Advances in Neural Information Processing Systems (NeurIPS), 2025

Bib

@inproceedings{tang2025physwin,
  title = {PhySwin: An Efficient and Physically-Informed Foundation Model for Multispectral Earth Observation},
  author = {Tang, Chong and Powell, J. and Koch, D. and Mullins, R. D. and Weddell, A. S. and Chauhan, J.},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2025},
}

EWSN

AdaTM: Logic-Inspired Adaptive Tsetlin Machines for Efficient and Effective Continual Learning on the Edge

Chong Tang, N. Singh, and J. Chauhan

In International Conference on Embedded Wireless Systems and Networks (EWSN), 2024

Bib

@inproceedings{tang2024adatm,
  title = {AdaTM: Logic-Inspired Adaptive Tsetlin Machines for Efficient and Effective Continual Learning on the Edge},
  author = {Tang, Chong and Singh, N. and Chauhan, J.},
  booktitle = {International Conference on Embedded Wireless Systems and Networks (EWSN)},
  year = {2024},
}