Yifan Hu 胡逸凡

I am an assistant professor in the Department of Statistics, Rutgers University (New Brunswick, NJ). I was a postdoc working with Prof. Daniel Kuhn from EPFL and Prof. Andreas Krause from ETH Zurich. Before that, I got my PhD in Operations Research from the University of Illinois at Urbana-Champaign on 2022, advised by Prof. Xin Chen and Prof. Niao He. I obtained a Bachelor in Mathematics from Nankai University.

My research interests lie in optimal decision-making under uncertainty, with an intersection of optimization, statistics, and operations research. I study problems arising from reinforcement learning, alignment and fine-tuning of LLM, large-scale causal inference and discovery, and operations management, leveraging methodologies from optimization and statistics, aiming to build new models and develop simple-to-implement methods with provable guarantees. See Research for details.

I am looking for motivated PhD students and research interns with backgrounds in mathematics, statistics, operations research, computer science or other related fields.

Feel free to send me an email if you find anything interesting. I am a skier, swimmer, hiker, and I play tennis.

yifan dot hu AT rutgers dot edu

Open Positions

If you are already a PhD student at Rutgers Stats, feel free to drop me an email.
For research intern, feel free to drop me an email with your CV and transcripts.
If you would like to apply for PhD at Rutgers Stats, please let me know and consider submitting your applications [Link].

News

Jul 2025. Paper. Our paper "Stochastic Optimization under Hidden Convexity" is accepted by SIAM Journal on Optimization.
Jun 2025. New Paper. Our paper "Statistical Inference for Conditional Group Distributionally Robust Optimization with Cross-Entropy Loss" is available online.
May 2025. New Paper. Our paper "MPO: An Efficient Post-Processing Framework for Mixing Diverse Preference Alignment" is accepted to ICML 2025.
Apr 2025. Blog. Our blog "Summary of Complexity Bounds for Algorithms in Minimization and Minimax Optimization" is accepted to ICLR Blogposts 2025. Check it out!
Jan 2025. Paper. Our paper "Global Group Fairness in Federated Learning via Function Tracking" got accepted to AISTATS 2025!
Dec 2024. Paper. Our paper "Causal Invariance Learning via Efficient Optimization of a Nonconvex Objective" is available online. Our optimization formulation provides a novel model for causal discovery and our method solves a nonconvex problem to global optimality efficiently, avoiding conducting an exhaustive search over all (exponentially many) subsets of covariates.
Oct 2024. Presentations.
- - I gave two guest lectures at the statistics department in Rutgers University, New Brunswick.
  - I was invited to give an oral presentation at the Cornell ORIE Young Researcher Workshop!
  - I gave two seminars in Rutgers University, New Brunswick.
  - I gave give a talk at INFORMS Annual Meeting in Seattle.
Sep 2024. Paper. Our new paper "Landscape of Policy Optimization for Finite Horizon MDPs with General State and Action" is available online. It established the benign nonconvex landscape of finite horizon MDP in control, inventory, and cash-balance problems, and the first sample complexity bounds for stochastic policy gradient methods for such problems.
Sep 2024. Paper. Three papers got accepted to NeurIPS 2024. See you in Vancouver!
Sep 2024. Talk. I gave a talk at OR2024 "Stochastic Hidden Convex Optimization in Network Revenue Management."
Aug 2024. Paper. Our paper "Multi-level Monte-Carlo Gradient Methods for Stochastic Optimization with Biased Oracles" is available online. This is the extended version of my earlier work in NeurIPS 2021.
Aug 2024. Paper. Our paper "Efficient Algorithms for A Class of Stochastic Hidden Convex Optimization and Its Applications in Network Revenue Management" got accepted by Operations Research.
Jul 2024. Talks. I gave two talks at International Symposium on Mathematical Programming in Montreal, Canada.
- - Stochastic Optimization under Hidden Convexity.
  - Three-stage Stochastic Programming is As Easy As Classical Stochastic Optimization.
Jun 2024. Paper. Our paper "Stochastic Bilevel Optimization with Lower-Level Contextual Markov Decision Processes" is online. First time being the last author!
May 2024. Papers. Two preprints are available online.
- - Stochastic Optimization Algorithms for Instrumental Variable Regression with Streaming Data.
  - Group Robust Preference Optimization in Reward-free Reinforcement Learning with Human Feedback.