Kun Xiang avatar

Kun Xiang (项鲲)

PhD Candidate @ Sun Yat-sen University  /  Research Intern @ Alibaba

Email  /  Github  /  LinkedIn  /  Google Scholar  /  ORCID  /  DBLP
About Me

Thanks for stopping by! 👋

I am a PhD candidate at Sun Yat-sen University, advised by Prof. Xiaodan Liang. My research agenda is Building the Reasoning Foundation of Physical AI: enabling multimodal systems to perceive, reason, model, and interact with the physical world.

I pursue this agenda through three core questions:

1) How can we elicit slow thinking and reliable supervision in multimodal models?

2) How can we build data-centric learning frameworks for scientific and physical intelligence?

3) How can we evaluate whether model reasoning is truly grounded in physical evidence?

News
  • [2026.06] One paper (Survey on Physical AI) accepted by TPAMI!
  • [2026.05] Our work (SeePhys Pro), a fine-grained modality-transfer benchmark for multimodal physics reasoning, is on arXiv!
  • [2026.03] Our workshop 3rd AI for Math Workshop: Toward Self-Evolving Scientific Agents is accepted to ICML 2026! See you in Seoul!
  • [2026.01] One paper (AtomThink) accepted by TPAMI!
  • [2025.10] Our work (AI4Physics), a survey to aligning perception, reasoning, modeling and interaction capabilities on physical AI, is on Arxiv!
  • [2025.09] One paper (SeePhys) accepted by NeurIPS 2025! See you in San Diego!
  • [2025.05] Code and Dataset of SeePhys have been released. Welcome to try!
  • [2025.05] Our work (SeePhys), a full spectrum multimodal benchmark for evaluating physics reasoning across different knowledge levels, is on Arxiv!
  • [2025.05] We organize a challenge for the 2nd AI for Math Workshop at ICML 2025. Welcome to try!
  • [2024.12] Our work (AtomThink), an o1 style reasoning framework via long CoT for complex multimodal mathematical tasks, is on Arxiv!
Selected Publications

Full publication list on Google Scholar. (* denotes equal contribution)

atomthink.png

AtomThink: Multimodal Slow Thinking with Atomic Step Reasoning

Kun Xiang*, Zhili Liu*, Terry Jingchen Zhang, Yinya Huang, and 6 more authors

Adaptive slow thinking for multimodal AI through atomic-step reasoning!

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2026

[arXiv] [Code] GitHub Repo stars
alignment.png

Aligning Perception, Reasoning, Modeling and Interaction: A Survey on Physical AI

Kun Xiang*, Terry Jingchen Zhang*, Yinya Huang*, Jixi He, and 12 more authors

A survey charting the path toward AI that understands and reasons about the physical world!

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2026

[arXiv] [Project Page]
seephys.png

SeePhys: Does Seeing Help Thinking? – Benchmarking Vision-Based Physics Reasoning

Kun Xiang*, Heng Li*, Terry Jingchen Zhang*, Yinya Huang*, and 10 more authors

A full spectrum multimodal benchmark for evaluating visual physics reasoning!

Neural Information Processing Systems (NeurIPS), 2025.

[arXiv] [Project Page] [Code] [Challenge (ICML Workshop)] [Dataset] GitHub Repo stars

EMOVA: Empowering Language Models to See, Hear and Speak with Vivid Emotions

Kai Chen*, Yunhao Gou*, Runhui Huang*, Zhili Liu*, Daxin Tan* and other 26 authors

Fully open-sourced Omni-modal LLMs with SoTA vision-language and speech abilities!

IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR), 2025

[arXiv] [Webpage] [Talk] [Talk (Chinese)] [Wechat Post] [Code] GitHub Repo stars
xiang2023toward.png

Toward robust diagnosis: A contour attention preserving adversarial defense for covid-19 detection

Kun Xiang, Xing Zhang, Jinwen She, Jinpeng Liu, and 3 more authors

Proceedings of the AAAI Conference on Artificial Intelligence, 2023

[arXiv] [Code]
xiang2021novel.png

A novel weight pruning strategy for light weight neural networks with application to the diagnosis of skin disease

Kun Xiang, Linlin Peng, Haiqiong Yang, Mingxin Li, and 3 more authors

Applied Soft Computing, 2021

[PDF]
seephyspro.png

SeePhys Pro: Diagnosing Modality Transfer and Blind-Training Effects in Multimodal RLVR for Physics Reasoning

Kun Xiang, Terry Jingchen Zhang, Zirong Liu, Bokai Zhou, and 14 more authors

A fine-grained modality-transfer benchmark and diagnostic suite for multimodal RLVR in physics reasoning!

arXiv preprint, 2026

[arXiv] [Project Page] [GitHub] [Dataset] [Challenge (ICML Workshop)]
geothinker.png

Thinking with Geometry: Active Geometry Integration for Spatial Reasoning

Haoyuan Li, Qihang Cao, Tao Tang, Kun Xiang, and 4 more authors

arXiv preprint, 2026

[arXiv] [Code] GitHub Repo stars
radar.png

RADAR: Revealing Asymmetric Development of Abilities in MLLM Pre-training

Yunshuang Nie, Bingqian Lin, Minzhe Niu, Kun Xiang, and 6 more authors

arXiv preprint, 2026

[arXiv] [Code]
Academic Services
Organizer:
  • 2nd/3st AI for Math Workshop @ ICML
Reviewer:
  • Conference: NeurIPS, ACL, AAAI
  • Journal: IEEE Transactions on Neural Networks and Learning Systems, IEEE Journal of Biomedical and Health Informatics, Information Sciences, Applied Soft Computing
Experiences
Alibaba DAMO Academy
May. 2026 - Present
Research Intern, working with Xin Li
Huawei Noah’s Ark Lab (Computer Vision group)
July. 2024 - May. 2026
Research Intern, working with Jianhua Han and Hang Xu
Sangfor Technologies Inc.
Apr. 2021 - Aug. 2021
Research Intern, working with Jingyan Jiang
Selected Awards

Outstanding Graduate of SYSU

2024

National Scholarship

2022

Postgraduate Scholarship of SYSU

2021