Zihao Li

PhD student, Electrical and Computer Engineering, Princeton University Email: zihaoli [at] princeton.edu

About me

Welcome to my homepage! I am currently a third-year PhD in ECE department at Princeton University. I am fortunate to be advised by Prof. Mengdi Wang. Before coming to Princeton, I obtained my B.S. in Mathematics and Applied Mathematics at Fudan University in 2022. You can find my CV here. My research interests lie in understanding modern machine learning from both theoretical and empirical perspectives. From the theoretical side, I aim to understand the principle of algorithms through tools in mathematics and statistics. From the empirical side, I aim to tackle impactful and challenging application problems through this understanding. My previous research includes topics such as reinforcement learning, generative models, and causal inference.

Miscellaneous

My name in Chinese is 李(=Li)子灏(=Zihao), which pronounces like Lee-Tzu-Hao in English. My pronouns are he/him/his. I am from Shanghai, China. I enjoy cooking and Total War when not working. My Erdos number is 5. You can find me through email or Wechat. All discussions are welcome!

Publications

• One-Layer Transformer Provably Learns One-Nearest Neighbor In Context
  Zihao Li, Yuan Cao, Cheng Gao, Yihan He, Mengdi Wang, Han Liu, Jason Matthew Klusowski, Jianqing Fan. Neurips 2024.

• Global Convergence in Training Large-Scale Transformers
  Cheng Gao, Yuan Cao, Zihao Li, Yihan He, Mengdi Wang, Han Liu, Jason Matthew Klusowski, Jianqing Fan. Neurips 2024.

• Theoretical Insights for Diffusion Guidance: A Case Study for Gaussian Mixture Models
  Yuchen Wu, Minshuo Chen, Zihao Li, Mengdi Wang, Yuting Wei. ICML 2024. [arXiv]

• Double Duality: Variational Primal-Dual Policy Optimization for Constrained Reinforcement Learning
  Zihao Li, Boyi Liu, Zhuoran Yang, Zhaoran Wang, Mengdi Wang. JMLR 2023. [arXiv]

• Policy Evaluation for Reinforcement Learning from Human Feedback: A Sample Complexity Analysis
  Zihao Li, Xiang Ji, Minshuo Chen, Mengdi Wang. AISTATS 2024. [arXiv]

• Provably efficient representation learning with tractable planning in low-rank pomdp
  Jiacheng Guo, Zihao Li,Huazheng Wang, Mengdi Wang, Zhuoran Yang, Xuezhou Zhang. ICML 2023. [arXiv]

• Fast Algorithms for Stackelberg Prediction Game with Least Squares Loss
  Jiali Wang, He Chen, Rujun Jiang, Xudong Li, Zihao Li. ICML 2021. [arXiv]

Preprints

• Diffusion Model for Data-Driven Black-Box Optimization
  Zihao Li, Hui Yuan, Kaixuan Huang, Chengzhuo Ni, Yinyu Ye, Minshuo Chen, Mengdi Wang. Submitted to Management Science. [arXiv]

• Regularized DeepIV with Model Selection
  Zihao Li, Hui Lan, Vasilis Syrkanis, Mengdi Wang, Masatoshi Uehara. [arXiv]

• Reinforcement Learning with Human Feedback: Learning Dynamic Choices via Pessimism
  Zihao Li, Zhuoran Yang, Mengdi Wang. [arXiv]