I am a tenure-track assistant professor in the Department of Biostatistics & Bioinformatics, Department of Computer Science, and Department of Electrical & Computer Engineering at Duke University. Prior to joining Duke, I was a Postdoctoral Scholar Research Associate in the Department of Computing and Mathematical Science at the California Institute of Technology. I received my Ph.D. degree in the Department of Computer Science at the University of California, Los Angeles. My research is centered around Machine Learning, with broad interests in the areas of Reinforcement Learning, Artificial Intelligence, Large Language Models, Optimization, High Dimensional Statistics, and their applications to real-world problems including Bioinformatics and Healthcare. My research goal is to develop computationally- and data-efficient machine learning algorithms with both strong empirical performance and theoretical guarantees.

[Prospective students] I am always happy to collaborate with highly motivated and self-driven students who are genuinely interested in our research. Before contacting me, please take the time to carefully read at least one of our lab’s papers (or a closely related paper) so that you have a clear understanding of the type of problems we study and the approaches we take. After doing so, please email me with: your CV (including coursework and technical skills) and a brief note (5–8 sentences) explaining: (1) which paper or research direction interests you, (2) why it interests you, and (3) what relevant background or experience you have. Due to the high volume of inquiries I receive, I am only able to respond to emails that demonstrate clear preparation and a strong, specific interest in our work.



Contact

Email: pan (DOT) xu (AT) duke (DOT) edu
Address: 2424 Erwin Road Room 9032, Durham, NC 27705
  ORCID:0000-0002-2559-8622
  Google Scholar


Recent News

  • [2026/01] Our papers on off-dynamics RL and heteroscedastic bandits get accepted to ICLR 2026.
    • MOBODY: Model-Based Off-Dynamics Offline Reinforcement Learning
    • Breaking the Total Variance Barrier: Sharp Sample Complexity for Linear Heteroscedastic Bandits with Fixed Action Set
  • [2025/12] Our paper on offline reinforcement learning for ICU decision support gets accepted to Critical Care Explorations.
    • rECMOmender: Offline Reinforcement Learning for Decision Support in Ventilator and Venovenous Extracorporeal Membrane Oxygenation Management
  • [2025/09] Our paper on distributionally robust MDPs with mixture models gets accepted to NeurIPS 2025.
    • Linear Mixture Distributionally Robust Markov Decision Processes
  • [2025/08] Our paper on generative AI for electronic health records (EHR) gets accepted to IEEE BHI 2025.
    • An Interactive Framework for Generating Clinical Data with Human Feedback
  • [2025/05] Our papers on distributionally robust MDPs with online interaction and linear robust MDPs with regularizations get accepted to ICML 2025.
    • Sample Complexity of Distributionally Robust Off-Dynamics Reinforcement Learning with Online Interaction
    • Robust Offline Reinforcement Learning with Linearly Structured $f$-Divergence Regularization
  • [2024/11] I am serving as an action editor for Transactions on Machine Learning Research (TMLR).
  • [2024/09] Our papers on pure exploration in batched bandits, randomized exploration in MARL, off-dynamics reinforcement learning, and distributionally robust RL get accepted to NeurIPS 2024.
    • Optimal Batched Best Arm Identification
    • Randomized Exploration in Cooperative Multi-Agent Reinforcement Learning
    • Off-Dynamics Reinforcement Learning via Domain Adaptation and Reward Augmented Imitation
    • Minimax Optimal and Computationally Efficient Algorithms for Distributionally Robust Offline Reinforcement Learning
  • [2024/09] My paper which extends my talk at New Faculty Highlights, focused on the efficient exploration and robust learning in decision making, gets published in AI Magazine.
    • Efficient and Robust Sequential Decision Making Algorithms
  • [2024/08] Our paper on self-supervised training of hypergraph neural networks gets accepted to TMLR.
    • PhyGCN: Pre-trained Hypergraph Convolutional Neural Networks with Self-supervised Learning
  • [2024/05] Our paper on exploration in deep reinforcement learning gets accepted to RLC 2024.
    • More Efficient Randomized Exploration for Reinforcement Learning via Approximate Sampling
  • [2024/05] Our paper on batched linear bandits gets accepted to ICML 2024.
    • Optimal Batched Linear Bandits
  • [2024/01] Our paper on distributionally robust reinforcement learning gets accepted to AISTATS 2024.
    • Distributionally Robust Off-Dynamics Reinforcement Learning: Provable Efficiency with Linear Function Approximation
  • [2024/01] Our paper on exploration in deep reinforcement learning gets accepted to ICLR 2024.
    • Provable and Practical: Efficient Exploration in Reinforcement Learning via Langevin Monte Carlo
  • [2023/12] Our paper on distributional robust bandits gets accepted to TMLR with a Featured Certification.
    • Wasserstein Distributionally Robust Policy Evaluation and Learning for Contextual Bandits
  • [2023/12] I have been invited to give a presentation at the New Faculty Highlights program at AAAI 2024.
  • [2023/12] Our paper on multi-agent multi-armed bandits gets accepted to AAAI 2024 as an Oral Presentation.
    • Finite-Time Frequentist Regret Bounds of Multi-Agent Thompson Sampling on Sparse Hypergraphs
  • [2023/08] Received an award from NSF on approximate sampling based exploration for sequential decision making. [NSF award information]
  • [2023/08] I am serving as an area chair for AISTATS 2024.
  • [2023/08] I am serving as an area chair for ICLR 2024.
  • [2023/04] Our paper on randomized exploration in multi-armed bandits gets accepted to ICML 2023.
    • Thompson Sampling with Less Exploration is Fast and Optimal
  • [2023/04] Our paper on Participatory AI gets accepted to FAccT 2023 and wins the Best Paper Award.
    • Queer In AI: A Case Study in Community-Led Participatory AI
  • [2023/03] I am serving as an area chair for NeurIPS 2023.
  • [2023/01] I received the Whitehead Scholar award from the Duke University School of Medicine.
  • [2023/01] Our paper on offline contextual bandits gets accepted to AISTATS 2023.
    • Distributionally Robust Policy Gradient for Offline Contextual Bandits
  • [2023/01] I am serving as an area chair for ICML 2023.
  • [2022/12] Our paper on multi-agent reinforcement learning gets accepted to ACM SIGMETRICS 2023.
    • Global Convergence of Localized Policy Iteration in Networked Multi-Agent Reinforcement Learning
  • [2022/09] Two papers on Thompson sampling and preference learning get accepted to NeurIPS 2022.
    • Finite-Time Regret of Thompson Sampling Algorithms for Exponential Family Multi-Armed Bandits
    • Active Ranking without Strong Stochastic Transitivity
  • [2022/08] I started at Duke University as an assistant professor.
  • [2022/01] I received the PIMCO Postdoctoral Fellowship in Data Science.
  • [2021/06] I received the UCLA Outstanding Graduate Student Research Award.
  • [2021/01] I am selected to the inaugural cohort of Rising Stars in Data Science by the Center for Data and Computing at the University of Chicago.
  • [2020/09] I am selected as a top reviewer for the International Conference on Machine Learning (ICML) 2020.
  • [2020/05] Our forecast on Combating COVID-19 has been adopted in the COVID-19 Forecasts by Centers for Disease Control and Prevention (CDC), California COVID Assessment Tool (CalCat) by California Department of Public Health (CDPH), and the COVID-19 Forecast Hub by the Reich Lab of the University of Massachusetts Amherst.
  • [2020/03] We launched the project Combating COVID-19 to model and predict the spread of coronavirus using machine learning empowered epidemic models. Details can be found at covid19.uclaml.org
  • [2020/02] One paper (P1) on rank aggregation is selected as Oral Presentation in AAAI 2020.
  • [2019/09] I am selected as a top reviewer for the Conference on Advances in Neural Information Processing Systems (NeurIPS) 2019.
  • [2019/06] One paper (P1) on reinforcement learning is selected as Oral Presentation in UAI 2019!
  • [2019/05] I gave a tutorial on "Nonconvex Optimization for Knowledge Discovery and Data Mining" with Quanquan Gu and Zhaoran Wang in SIAM International Conference on Data Mining (SDM'19). Slides can be found at [Slides]
  • [2018/09] Two papers (P1, P2) on nonconvex optimization are selected as Spotlight in NeurIPS 2018!
  • [2018/05] One paper (P1) on high-dimensional statistics is selected as Long Oral Talk in ICML 2018.
  • [2016/06] I received the Presidential Fellowship in Data Science from the University of Virginia.