💫 About Me
Hi there!👋
Welcome to the homepage of Yushu Li (李昱澍)!
I am a Ph.D. student in Electrical and Computer Engineering at the University of British Columbia (UBC) and a Ph.D. Student Researcher at the Vector Institute. At UBC, I am a member of the Trusted and Efficient AI (TEA) Lab, advised by Prof. Xiaoxiao Li. I am currently an Applied Scientist Intern at Amazon Web Services (AWS), working on robust and efficient reinforcement learning (RL).
Previously, I obtained my M.Eng. in Information and Communication Engineering from the South China University of Technology (SCUT), supervised by Prof. Kui Jia and co-supervised by Dr. Xun Xu. I was also a visiting student at the Institute for Infocomm Research (I2R), A*STAR, Singapore, working with Dr. Xun Xu and Dr. Xulei Yang. I received my B.Eng. from the School of Electronic and Information Engineering, SCUT.
My research focuses on Large Language Models, Reinforcement Learning, LLM Agents, and Test-Time Adaptation. I am open to collaboration with researchers and engineers across academia and industry.
📫 Feel free to reach out: yushul@student.ubc.ca
📄 My resume can be found here.
🔥 News
- 2026.08: 🥳 Our paper, Exploiting Vision Language Model for Training-Free 3D Point Cloud Understanding via Improved Graph Score Propagation, has been accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).
- 2026.07: ⛵️ Started as an Applied Scientist Intern at Amazon Web Services (AWS) in Cupertino, working on robust and efficient reinforcement learning (RL).
- 2026.05: 🥳 Two papers have been accepted at ICML 2026, including work on stable agentic reinforcement learning and retrieval-augmented LVLMs.
- 2026.05: ✨ Our work on mitigating reward hacking through directional alignment is accepted at ICML 2026 AIWild.
- 2026.03: ✨ Our new preprint, Spend Less, Reason Better: Budget-Aware Value Tree Search for LLM Agents, is now available online.
- 2026.01: 🥳 Our work on Token Hidden Reward has been accepted at ICLR 2026.
- 2025.09: 🏅 Awarded the Four Year Fellowship at UBC.
- 2025.09: ⛵️ Joined the Trusted and Efficient AI (TEA) Lab at UBC and became a Ph.D. Student Researcher at the Vector Institute.
- 2025.07: 🎉 One paper has been accepted at ICCV 2025. Congratulations to Tiankai!
- 2025.01: 🥳 Two papers have been accepted at ICLR 2025.
- 2025.01: 🎉 I am serving as a reviewer for ICME 2025 and TMLR.
- 2024.12: 🥳 Our work on Exploring Human-in-the-Loop Test-Time Adaptation by Synergizing Active Learning and Model Selection is accepted by Transactions on Machine Learning Research (TMLR).
- 2024.12: ✨ Our new preprint titled Efficient and Context-Aware Label Propagation for Zero-/Few-Shot Training-Free Adaptation of Vision-Language Model is now available online.
- 2024.12: 🎉 I am serving as a reviewer for IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).
- 2024.12: 🎉 I will be serving as a reviewer for ICML 2025.
- 2024.10: ✨ Our new preprint titled On the Adversarial Risk of Test Time Adaptation: An Investigation into Realistic Test-Time Data Poisoning is now available online.
- 2024.09: 🎓 I have completed my SIPGA program at I2R, A*STAR, Singapore.
- 2024.08：🎉 I will be serving as a reviewer for ICLR 2025.
- 2024.05: ✨ New preprint on Exploring Human-in-the-Loop Test-Time Adaptation by Synergizing Active Learning and Model Selection is available on arXiv.
- 2024.03: ⛵️ Starting my visiting student program “Singapore Inernational Pre-graduate Award (SIPGA)” at I2R, A*STAR, Singapore.
- 2023.07: 🥳 Our work on On the Robustness of Open-World Test-Time Training: Self-Training with Dynamic Prototype Expansion is accepted by ICCV 2023 as Oral presentation (1.8% acceptance rate).
📝 Publications
💥Exploiting Vision Language Model for Training-Free 3D Point Cloud Understanding via Improved Graph Score Propagation
Tiankai Chen, Yushu Li, Adam Goodge, Fei Teng, Xulei Yang, Zhen Liang, Tianrui Li, Xun Xu
IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2026.
💥Spend Less, Reason Better: Budget-Aware Value Tree Search for LLM Agents
Yushu Li*, Wenlong Deng*, Jiajin Li, Xiaoxiao Li
arXiv preprint, 2026.
💥On Group Relative Policy Optimization Collapse in Agent Search: The Lazy Likelihood-Displacement
Wenlong Deng*, Yushu Li*, Boying Gong*, Yi Ren, Christos Thrampoulidis, Xiaoxiao Li
International Conference on Machine Learning (ICML), 2026.
💥When RAG Hurts: Diagnosing and Mitigating Attention Distraction in Retrieval-Augmented LVLMs
Beidi Zhao, Wenlong Deng, Xinting Liao, Yushu Li, Nazim Shaikh, Yao Nie, Xiaoxiao Li
International Conference on Machine Learning (ICML), 2026.
💥Directional Alignment Mitigates Reward Hacking in Reinforcement Learning for Language Models
Wenlong Deng, Jiaji Huang, Kaan Ozkara, Yushu Li, Christos Thrampoulidis, Xiaoxiao Li, Youngsuk Park
ICML 2026 AIWild, 2026.
💥Token Hidden Reward: Steering Exploration-Exploitation in Group Relative Deep Reinforcement Learning
Wenlong Deng, Yi Ren, Yushu Li, Boying Gong, Danica J. Sutherland, Xiaoxiao Li, Christos Thrampoulidis
International Conference on Learning Representations (ICLR), 2026.
💥Exploiting Vision Language Model for Training-Free 3D Point Cloud OOD Detection via Graph Score Propagation
Tiankai Chen, Yushu Li, Adam Goodge, Tianrui Li, Fei Teng, Xulei Yang, Xun Xu
IEEE/CVF International Conference on Computer Vision (ICCV), 2025.
💥Efficient and Context-Aware Label Propagation for Zero-/Few-Shot Training-Free Adaptation of Vision-Language Model
Yushu Li*, Yongyi Su*, Adam Goodge, Kui Jia, Xun Xu
International Conference on Learning Representations (ICLR), 2025.
💥On the Adversarial Risk of Test Time Adaptation: An Investigation into Realistic Test-Time Data Poisoning
Yongyi Su*, Yushu Li*, Kui Jia, Chuan-Sheng Foo, Xun Xu
International Conference on Learning Representations (ICLR), 2025.
Exploring Human-in-the-Loop Test-Time Adaptation by Synergizing Active Learning and Model Selection
Yushu Li*, Yongyi Su*, Xulei Yang, Kui Jia, Xun Xu
Transactions on Machine Learning Research (TMLR), 2024.
On the Robustness of Open-World Test-Time Training: Self-Training with Dynamic Prototype Expansion
Yushu Li, Xun Xu, Yongyi Su and Kui Jia
IEEE/CVF International Conference on Computer Vision (ICCV), 2023
(Oral presentation 1.8% acceptance rate)
Dynamic Shuffle: An Efficient Channel Mixture Method
Kaijun Gong, Zhuowen Yin, Yushu Li, Kailing Guo, Xiangmin Xu
Preprint, 2023.
💻 Internships
- 2026.07.20 - now, Applied Scientist Intern, Amazon Web Services (AWS), Cupertino, USA. Working on robust and efficient reinforcement learning (RL).
- 2024.03 - 2024.09, Visiting Student, Institute for Infocomm Research (I2R), Agency for Science, Technology and Research (A*STAR), Singapore. Mentor: Drs. Xun Xu and Xulei Yang.
🎖 Honors and Awards
- Four Year Fellowship, The University of British Columbia (2025-2029)
- University-level Scholarship, South China University of Technology (2019-2025)
🎓 Education
- 2025.09 - now, Ph.D. in Electrical and Computer Engineering (ECE), The University of British Columbia (UBC), Canada
- 2022.09 - 2025.06, M.Eng. in Information and Communication Engineering, South China University of Technology (SCUT), China
- 2018.09 - 2022.06, B.Eng. in Information Engineering, South China University of Technology (SCUT), China
📖 Academic Service
- Conference Reviewer: ACM MM, NeurIPS, ICLR, ICML, ICME, CVPR
- Journal Reviewer: TIP, TMLR, TPAMI
🛠 Skills
- Programming: Python, PyTorch, C/C++, MATLAB, LaTeX
- Languages: Native Chinese speaker; fluent in English
This page was last modified by Yushu Li on: