I am a fifth-year PhD candidate in Computer Sciences at the University of Wisconsin–Madison. My research lies at the intersection of optimization and machine learning theory. I study the geometric and statistical structures that make challenging learning and optimization problems efficiently solvable, and design algorithms that exploit these structures. I am very fortunate to be advised by Prof. Jelena Diakonikolas and to have collaborated with Prof. Ilias Diakonikolas and many other wonderful researchers. Before coming to Madison, I earned my B.S. in Mathematics from Shandong University.

I am on the job market for fall 2027 -- seeking postdoctoral and research positions in academia or industry!

Publications

1. Learning Positive Combinations of ReLUs in Fully Polynomial Time

   Ilias Diakonikolas, Jelena Diakonikolas, Puqian Wang, Nikos Zarifis (alphabetical)

   In submission · 2026 ECCC, arXiv

2. Efficient Optimization on the Euclidean Sphere: Riemannian Gradient Alignment as a Unifying Principle

   Puqian Wang, Nikos Zarifis, Jelena Diakonikolas

   In submission · 2026 arXiv

3. ARO: A New Lens on Matrix Optimization for Large Models

   Wenbo Gong, Javier Zazo, Qijun Luo, Puqian Wang, James Hensman, Chao Ma

   In submission · 2026 arXiv

4. Robustly Learning Monotone Single-Index Models

   Puqian Wang^*, Nikos Zarifis^*, Ilias Diakonikolas, Jelena Diakonikolas

   NeurIPS 2025 arXiv

5. Robustly Learning Monotone Generalized Linear Models via Data Augmentation

   Nikos Zarifis^*, Puqian Wang^*, Ilias Diakonikolas, Jelena Diakonikolas

   COLT 2025 arXiv

6. Sample and Computationally Efficient Robust Learning of Gaussian Single-Index Models

   Puqian Wang, Nikos Zarifis, Ilias Diakonikolas, Jelena Diakonikolas

   NeurIPS 2024 arXiv

7. Robustly Learning Single-Index Models via Alignment Sharpness

   Nikos Zarifis^*, Puqian Wang^*, Ilias Diakonikolas, Jelena Diakonikolas

   ICML 2024 arXiv

8. Near-Optimal Bounds for Learning Gaussian Halfspaces with Random Classification Noise

   Ilias Diakonikolas, Jelena Diakonikolas, Daniel Kane, Puqian Wang, Nikos Zarifis (alphabetical)

   NeurIPS 2023 arXiv

9. Information-Computation Tradeoffs for Learning Margin Halfspaces with Random Classification Noise

   Ilias Diakonikolas, Jelena Diakonikolas, Puqian Wang, Nikos Zarifis (alphabetical)

   COLT 2023 arXiv

10. Robustly Learning a Single Neuron via Sharpness

    Puqian Wang^*, Nikos Zarifis^*, Ilias Diakonikolas, Jelena Diakonikolas

    ICML 2023, Oral Presentation arXiv

11. Potential Function-based Framework for Making the Gradients Small in Convex and Min-Max Optimization

    Jelena Diakonikolas, Puqian Wang (alphabetical)

    SIAM Journal on Optimization 2022 arXiv

^* Equal contribution.

Talks

• When Is Nonconvexity Not a Problem -- Benign Landscapes and Hidden Curvature in Learning on the Sphere, MOPTA, August 2026, Lehigh University
• Robustly Learning Single-Index Models, INFORMS Annual Meeting, October 2025, Atlanta, GA
• Sample and Computationally Efficient Robust Learning of Gaussian Single-Index Models, IFDS Forum, October 2024, UW-Madison
• Robustly Learning a Single Neuron via Alignment Sharpness, 60th Allerton Conference, September 2024, Urbana-Champaign, IL
• Robustly Learning a Single Neuron via Alignment Sharpness, International Symposium on Mathematical Programming, July 2024, Montreal, Canada

Experience

Research Intern, Microsoft Research Cambridge, July-September 2025.

Teaching

Teaching Assistant, UW-Madison

• Fall 2021 CS577 Introduction to Algorithms
• Fall 2022 CS726 Nonlinear Optimization I