Bio

I am Fan LIU, a research student at HKUST(GZ). My current interest is in trustworthy LLM, LLM for Reasoning and LLM-Powered Agents, etc. My research papers have been published in NeurIPS/KDD/ICMLW/TFS, etc. For more details, please refer to [Google Scholar]

Currently, I am particularly interested in building trustworthy and effective LLMs, which I approach through three key aspects: evaluation, foundational capabilities, and real-world applications.

  1. LLM Evaluation: Understanding the boundaries of LLMs, including their trustworthiness and reasoning capabilities. This involves assessing robustness, reliability, and alignment with human values.

  2. Enhancing Fundamental Capabilities: Based on evaluation insights, I explore methods to improve LLM trustworthiness and reasoning, including defending against jailbreak attacks, ensuring alignment with human values, and addressing fundamental challenges in logical reasoning and reflection.

  3. LLM-Powered Agents and Applications: Leveraging improved foundational capabilities to develop trustworthy, real-world AI agents. This involves studying agent alignment, decision-making, and adaptive interactions in dynamic environments.

Through this structured approach, my research aims to bridge the gap between LLM evaluation, theoretical advancements, and practical deployment, ultimately contributing to the development of more reliable and capable AI systems.

If you are interested in my research, feel free to reach out for discussions, collaborations, internship opportunities, or any related inquiries.

Email: liufanuestc AT DOT com

News: Safe Intelligence AI unveils a revolutionary initiative to enhance the reliability of Large Language Models! Our new project Safe Intelligence AI (安智 AI) is dedicated to establishing the trustworthiness of LLMs throughout the entire development process. For more information, please visit our Website.

Recent Works

  • [Arxiv] Fan LIU, Wenshuo Chao, Naiqiang Tan, Hao Liu, Bag of Tricks for Inference-time Computation of LLM Reasoning, Arxiv, 2025. [pdf], [Code]
  • [WWW] Fan LIU, Hao Liu, Subgraph Federated Unlearning, WWW, 2025.
  • [Arxiv] Fan LIU, Yue Feng, Zhao Xu, Lixin Su, Xinyu Ma, Dawei Yin, Hao Liu, JAILJUDGE: A Comprehensive Jailbreak Judge Benchmark with Multi-Agent Enhanced Explanation Evaluation Framework, Arxiv, 2024. [Project Page],[pdf], [Code], [Dataset], [Model],[Coverage] 🔥🚀 Model 400+ Downloads
  • [NeurIPS] Zhao Xu, Fan LIU, Hao Liu, Bag of Tricks: Benchmarking of Jailbreak Attacks on LLMs, NeurIPS D&B, 2024. [pdf], [Code],[Coverage]
  • [Arxiv] Fan LIU, Zhao Xu, Hao Liu, Adversarial Tuning: Defending Against Jailbreak Attacks for LLMs, Arxiv, 2024. [pdf]
  • [Arxiv] Fan LIU, Siqi Lai, Yansong Ning, Hao Liu, Bkd-FedGNN: A Benchmark for Classification Backdoor Attacks on Federated Graph Neural Network, Arxiv, 2023. [pdf], [Code]
  • [KDD] Fan LIU, Weijia Zhang, Hao Liu, Robust Spatiotemporal Traffic Forecasting with Reinforced Dynamic Adversarial Training, KDD, 2023.
  • [NeurIPS] Fan LIU, Hao Liu, Wenzhao Jiang, Practical Adversarial Attacks on Spatiotemporal Traffic Forecasting Models, NeurIPS, 2022. [Blog], [Code]

Education and Experience

  • 2022: Graduate student at HKUST(GZ)
  • 2021: Intern at HKUST(GZ)
  • 2020: Intern at MSRA (StarBridge Program)
  • 2020: B.S. from UESTC
  • 2019: Research visit at UBC

Awards, Acknowledgements, and Services

  • Reviewer for Conference: ICLR 2024-2025, NeurIPS 2023-2024, KDD 2023-2025, WWW 2025, AISTATS 2025, AdvML-Frontiers (ICML 2023 Workshop), FL4Data-Mining (KDD 2023 Workshop)
  • Reviewer for Journal: ITS, Transactions On SMC: Systems, Physica A, TFS, TII
  • TPC member: FL4Data-Mining (KDD 2023 Workshop)
  • KDD Student Travel Award (2023)
  • RBM Student Travel Grant (2023)
  • Outstanding Undergraduate Thesis Award
  • Outstanding Undergraduate Student
  • Excellent Student Scholarship (2017-2020)