Scholar
Zihao Li
Google Scholar ID: tbu1jqUAAAAJ
Princeton University
Citations & Impact
Citations
309
H-index
8
i10-index
6
Publications
11
Co-authors
10
list available
Contact
Publications
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
- One-Layer Transformer Provably Learns One-Nearest Neighbor In Context, Neurips 2024.
- Global Convergence in Training Large-Scale Transformers, Neurips 2024.
- Theoretical Insights for Diffusion Guidance: A Case Study for Gaussian Mixture Models, ICML 2024.
- Double Duality: Variational Primal-Dual Policy Optimization for Constrained Reinforcement Learning, JMLR 2023.
- Policy Evaluation for Reinforcement Learning from Human Feedback: A Sample Complexity Analysis, AISTATS 2024.
- Provably efficient representation learning with tractable planning in low-rank pomdp, ICML 2023.
- Fast Algorithms for Stackelberg Prediction Game with Least Squares Loss, ICML 2021.
Research Experience
- No specific job positions or work experience mentioned.
Education
- Received B.S. in Mathematics and Applied Mathematics from Fudan University in 2022; currently pursuing a Ph.D. at Princeton University, advised by Prof. Mengdi Wang.
Background
- Currently a third-year PhD student in the ECE department at Princeton University. Research interests lie in understanding modern machine learning from both theoretical and empirical perspectives. Theoretically, aims to understand the principle of algorithms through tools in mathematics and statistics. Empirically, aims to tackle impactful and challenging application problems through this understanding. Previous research includes topics such as reinforcement learning, generative models, and causal inference.
Miscellany
- From Shanghai, China. Enjoys cooking and playing Total War when not working. His Erdos number is 5. Can be reached via email or Wechat. All discussions are welcome!
