Junchi Yang

I am a postdoctoral researcher at Argonne National Laboratory, working with Feng Qiu. I completed my Ph.D. in Computer Science at ETH Zurich, under the supervision of Niao He. My research interests lie in optimization and its applications in machine learning and power systems. Prior to this, I received a master's degree in Industrial Engineering from the University of Illinois Urbana-Champaign. I received a bachelor's degree in Applied Mathematics and Economics from UCLA. Email: junchi.yang (at) anl (dot) gov Linkedin. Google Scholar.

News

• [2024.2] I am attending the 2024 Conference on Innovative Smart Grid Technologies (ISGT) in February and the INFORMS Optimization Society Conference in March.

• [2024.2] My Ph.D. Thesis “Towards Near-Optimal and Adaptive Algorithms in Minimax Optimization” is now available online. Check it out.

• [2024.1] Our paper “Parameter-Agnostic Optimization under Relaxed Smoothness” is accepted to AISTATS 2024.

• [2024.1] I started as a postdoctoral researcher at Argonne National Laboratory.

• [2023.09] Two co-first-authored papers are accepted to NeurIPS 2023. One of them is selected to be a spotlight paper.

• [2023.09] I've successfully defended my PhD thesis. Grateful for this academic journey.

• [2023.06] I am giving a talk at the 20th EUROpt Workshop in Budapest in August.

• [2023.05] Our paper “Two Sides of One Coin: the Limits of Untuned SGD and the Power of Adaptive Methods” is online.

• [2023.03] I am giving a talk at the SIAM Conference on Optimization (OP23) in Seattle in June.

• [2023.01] Our paper “TiAda: A Time-scale Adaptive Algorithm For Nonconvex Minimax Optimization” is accepted by ICLR 2023.

Publications

(* indicates equal contribution)

• Parameter-Agnostic Optimization under Relaxed Smoothness
  Florian Hübler, Junchi Yang, Xiang Li, Niao He.
  AISTATS 2024 (preliminary version in NeurIPS OPT Workshop 2023) [arXiv]

• Optimal Guarantees for Algorithmic Reproducibility and Gradient Complexity in Convex Optimization
  Liang Zhang*, Junchi Yang*, Amin Karbasi, Niao He.
  NeurIPS 2023 (spotlight). [arXiv]

• Two Sides of One Coin: the Limits of Untuned SGD and the Power of Adaptive Methods
  Junchi Yang*, Xiang Li*, Ilyas Fatkhullin, and Niao He.
  NeurIPS 2023. [arXiv]

• TiAda: A Time-Scale Adaptive Algorithm For Nonconvex Minimax Optimization
  Xiang Li, Junchi Yang, and Niao He.
  ICLR 2023 (preliminary version in NeurIPS OPT Workshop 2022) [arXiv] [ICLR]

• Nest Your Adaptive Algorithm for Parameter-Agnostic Nonconvex Minimax Optimization
  Junchi Yang*, Xiang Li*, and Niao He.
  NeurIPS 2022 [arXiv] [NeurIPS]

• Faster Single-Loop Algorithms for Minimax Optimization without Strong Concavity
  Junchi Yang, Antonio Orvieto, Aurelien Lucchi, and Niao He.
  AISTATS 2022 [arXiv] [AISTATS]

• The Complexity of Nonconvex-Strongly-Concave Minimax Optimization
  Siqi Zhang*, Junchi Yang*, Cristobal Guzman, Negar Kiyavash and Niao He.
  UAI 2021 [arXiv] [UAI]

• A Catalyst Framework for Minimax Optimization
  Junchi Yang, Siqi Zhang, Negar Kiyavash, and Niao He.
  NeurIPS 2020 [NeurIPS]

• Global Convergence and Variance Reduction for a Class of Nonconvex-Nonconcave Minimax Problems
  Junchi Yang, Negar Kiyavash, and Niao He.
  NeurIPS 2020 [arXiv] [NeurIPS]

Talks

• 20th EUROpt Workshop 2023. Title: “From SGD to Adaptive Methods: Benefits of Adaptive Gradient Techniques”.

• SIAM Conference on Optimization 2023 (OP23). Title: “Adaptive Algorithms for Nonconvex Minimax Optimization”.

• CSL Student Conference 2021. Title: “A Catalyst Framework for Minimax Optimization”.

• INFORMS 2020 Annual Meeting. Title: “Simple and Efficient Algorithms for Classes of Nonconvex Minimax Optimization”.

Teaching

• Optimization for Data Science. ETH Zurich. 2022 Spring, 2023 Winter.

• Foundations of Reinforcement Learning. ETH Zurich. 2021 Fall.

• Analysis of Data. UIUC. 2019 Fall.

Service

Reviewer for NeurIPS, ICML, ICLR, AISTATS and TMLR.

Miscellaneous

• I received ISE Student Fellowship at UIUC 2017-2018.

• Top reviewer for NeurIPS 2022.

• I play volleyball and used to play badminton and table tennis. I am learning snowboarding and tennis.