Journal Submission under Review

[1] Efficient Algorithms for Minimizing Compositions of Convex Functions and Random Functions and Its Applications in Network Revenue Management. [ArXiv] [Slides]

(Alphabetical) Xin Chen, Niao He, Yifan Hu, and Zikun Ye. 2022. 

• Minor revision at Operations Research.

• Selected as an oral presentation at MSOM Supply Chain SIG 2022.

• Selected as an oral presentation at RMP Conference 2022. 

[2] Stochastic Optimization under Hidden Convexity. [ArXiv] [Slides]

(Alphabetical) Ilyas Fatkhullin, Niao He, Yifan Hu. 2023. 

• Preliminary version in NeurIPS OPT for Machine Learning Workshop 2023. [Link]

• Journal version submitted.

Journal Publication

[3] Sample Complexity of Sample Average Approximation for Conditional Stochastic Optimization. [Link]

Yifan Hu, Xin Chen, and Niao He. SIAM Journal on Optimization 2020.

Invited Book Chapter

[4] Stochastic Biased Gradient Methods. 

Yifan Hu. Accepted by Encyclopedia of Optimization, 3rd edition. Forthcoming.

Conference Proceedings (*equal contribution)

[5] Contextual Stochastic Bilevel Optimization. [Link] [Slides]

Yifan Hu, Jie Wang, Yao Xie, Andreas Krause, Daniel Kuhn. NeurIPS 2023.

[6] On the Bias-Variance-Cost Tradeoff of Biased Stochastic Optimization. [Link]

Yifan Hu, Xin Chen, and Niao He. NeurIPS 2021. 

[7] Biased Stochastic First-order Methods for Conditional Stochastic Optimization and Its Applications in Meta Learning. [Link]

Yifan Hu*, Siqi Zhang*, Xin Chen, and Niao He. NeurIPS 2020.

[8] Generalization Bounds of Nonconvex-(Strongly)-Concave Stochastic Minimax Optimization. [Link]

Siqi Zhang*, Yifan Hu*, Liang Zhang, Niao He. AISTATS 2024.

• Preliminary version in NeurIPS OPT for Machine Learning Workshop 2022.

[9] Distributionally Robust Model-based Reinforcement Learning with Large State Spaces. [Link]

Shyam Sundhar Ramesh, Pier Giuseppe Sessa, Yifan Hu, Andreas Krause, Ilija Bogunovic. AISTATS 2024.

• Preliminary version in NeurIPS ReALML Workshop 2023. 

Preprints

[10] Stochastic Bilevel Optimization with Lower-Level Contextual Markov Decision Processes. [ArXiv]

Vinzenz Thoma, Barna Pasztor, Andreas Krause, Giorgia Ramponi, Yifan Hu.

• Preliminary version in Agentic Markets Workshop of ICML, 2024.

• TL;DR: How to design environments of MDPs for multiple players to achieve robustness, exploration, imitation, and personalization.

[11] Stochastic Optimization Algorithms for Instrumental Variable Regression with Streaming Data. [ArXiv]

Xuxing Chen, Abhishek Roy, Yifan Hu, Krishna Balasubramanian.

• TL;DR: Formulating instrumental variable regression as optimization, you can avoid the first stage estimation in the two stage regression and come up with fully online algorithm with cheap per-iteration costs.

[12] Group Robust Preference Optimization in Reward-free Reinforcement Learning with Human Feedback. [ArXiv]

Shyam Sundhar Ramesh, Yifan Hu, Iason Chaimalas, Viraj Mehta, Pier Giuseppe Sessa, Haitham Bou Ammar, Ilija Bogunovic. 

• Preliminary version in ELLIS Robust LLMs Workshop, 2024.

• TL;DR: When there are multiple language data sources, how to fine-tune a language model that performs well on all of them.

Technical Notes

Nonconvex Variance Reduction: General Analysis, Biased Oracle and Optimal Algorithms for Stochastic Bilevel Optimization.

Liang Zhang, Yifan Hu, and Niao He. 2021. 

Presentations

• Contextual Stochastic Bilevel Optimization.

  • • INFORMS Optimization Society Conference, Mar 2024.

    • INFORMS Annual Meeting, Oct 2023.

    • International Conference on Stochastic Programming, Jul 2023.

    • SIAM Conference on Optimization, Seattle, May 2022.

• Global Converging Algorithms for Nonconvex Network Revenue Management via (Inaccessible) Hidden Convexity.

  • • INFORMS Annual Meeting, Oct 2023.

    • Swiss Operations Research Day, Jun 2023.

    • Seminar at University of Zurich, May 2023.

    • INFORMS Annual Meeting, Oct 2022.

    • Seminar of Data Science Lab, MIT, Sep 2022.

    • ICCOPT, Jul 2022.

    • PhD Seminar, Institute of Machine Learning, ETH Zurich, Apr 2022.

    • RAO Lab Seminar, EPFL Mar 2022.

• On the Bias-Variance-Cost Tradeoff of Gradient Methods for Stochastic Optimization with Biased Oracles.

  • • NeurIPS 2021, Dec 2021.

    • Seminar at Nanjing University, Nov 2021.

    • PhD Seminar, Institute of Machine Learning, ETH Zurich, Oct 2021.

    • INFORMS Annual Meeting, Oct 2021.

• Biased First-Order Methods for Conditional Stochastic Optimization and Its Applications in Meta-Learning.

  • • Antai College of Economics and Management, Shanghai Jiaotong University, May 2021.

    • NeurIPS 2020, Dec 2020.

    • INFORMS Annual Meeting, Oct 2020.

    • CSL Student Conference, Feb 2020.