Jiancong Xiao

prof_pic.jpg

Welcome to my homepage! I am a tenure-track Assistant Professor in the Department of Computer Science at the National University of Singapore (NUS).

Previously, I was a postdoctoral researcher at the University of Pennsylvania, where I worked with Prof. Qi Long and Prof. Weijie J. Su. I obtained my Ph.D. from The Chinese University of Hong Kong, Shenzhen, where I was advised by Prof. Zhi-Quan (Tom) Luo and worked closely with Prof. Ruoyu Sun. Before starting my doctoral studies, I spent two years working in finance. I received my M.S. degree from The Chinese University of Hong Kong and my Bacholar’s degree from Sun Yat-sen University.

Research Interest: I am broadly interested in learning theory, statistics and optimization, with a focus on establishing foundations for responsible and trustworthy machine learning models. My recent research focuses on mathematical and statistical topics in large language models (LLMs).

  1. Theoretical Foundations for LLM: Analyzing algorithmic biases in preference alignment methods, developing preference alignment algorithms, exploring game-theoretic approaches to alignment, and investigating alignment from the perspective of social choice theory.

  2. LLM Training and Fine-Tuning: Establishing theoretical foundations for fine-tuning and post-training algorithms. Addressing calibration issues in LLMs.

  3. Adversarial Robustness: Developing theoretical frameworks for understanding adversarial examples, robust overfitting, and adversarially robust generalization.

  4. Classical Learning Theory: Studying Generalization (Rademacher complexity, Uniform stability, Pac-Bayes), Optimization (non-convex, non-smooth problem).

[Announcement] I am actively looking for PhD students for Fall 2027, as well as research assistants and research interns. If you are interested, please feel free to reach out via email.

News

Jul 09, 2026 Our paper titled “Beyond Post-Hoc Temperature Scaling: Bilevel Optimization for LLM Calibration” has been accepted to COLM. Thanks for all the collaborators!
Jul 01, 2026 I officially joined NUS SoC as an Assistant Professor.
Jun 24, 2026 Our paper, Adversarial Rademacher Complexity of Deep Neural Networks, has been accepted for publication in JMLR. The paper has had a rather eventful journey. It was first submitted to ICLR 2022 and received scores of 8/8/8/6, but was not accepted. We subsequently submitted a revised version to JMLR, where the review process took longer than we had anticipated. We are very glad that the journey has now reached a positive conclusion. I am deeply grateful to everyone who supported and helped me along the way.
Apr 27, 2026 Our paper titled “Statistical Impossibility and Possibility of Aligning LLMs with Human Preferences: From Condorcet Paradox to Nash Equilibrium” has been accepted to AoS. Thanks for all the collaborators!
Aug 18, 2025 Our paper titled “On the Algorithmic Bias of Aligning Large Language Models with RLHF: Preference Collapse and Matching Regularization” has been accepted to JASA. Thanks for all the collaborators!

Selected Publications

  1. ICML 2025
    Restoring Calibration for Aligned Large Language Models: A Calibration-Aware Fine-Tuning Approach
    Jiancong Xiao, Bojian Hou, Zhanliang Wang, Ruochen Jin, Qi Long, Weijie J. Su, and Li Shen
    In Proceedings of the 42nd International Conference on Machine Learning, 2025
  2. Annals of Statistics
    Statistical Impossibility and Possibility of Aligning LLMs with Human Preferences: From Condorcet Paradox to Nash Equilibrium
    Kaizhao Liu, Qi Long, Zhekun Shi, Weijie J. Su, and Jiancong Xiao
    Annals of Statistics (AoS) , 2026
  3. JASA
    On the Algorithmic Bias of Aligning Large Language Models with RLHF: Preference Collapse and Matching Regularization
    Jiancong Xiao, Ziniu Li, Xingyu Xie, Emily Getzen, Cong Fang, Qi Long, and Weijie J. Su
    Journal of the American Statistical Association (JASA) , 2025
  4. COLT 2024
    Bridging the Gap: Rademacher Complexity in Robust and Standard Generalization
    Jiancong Xiao, Ruoyu Sun, Qi Long, and Weijie J. Su
    In Proceedings of the Thirty Seventh Conference on Learning Theory, 2024
  5. NeurIPS 2022 Spotlight
    Stability Analysis and Generalization Bounds of Adversarial Training
    Jiancong Xiao, Yanbo Fan, Ruoyu Sun, Jue Wang, and Zhi-Quan Luo
    In Advances in Neural Information Processing Systems, 2022