Research

My research interests include stochastic control, reinforcement learning, nonconvex and stochastic optimization, diffusion models and applications in management and finance.

• Yinbin Han, Meisam Razaviyayn, and Renyuan Xu. Fine-tuning diffusion models: A stochastic control approach. Preprint, 2024.

• Yinbin Han, Meisam Razaviyayn, and Renyuan Xu. Neural Network-Based Score Estimation in Diffusion Models: Optimization and Generalization. International Conference on Learning Representations (ICLR), 2024.
  • Short version accepted by NeurIPS Workshop on Diffusion Models, 2023.
  • Long version to be submitted to Mathematics of Operations Research.
• Yinbin Han, Meisam Razaviyayn, and Renyuan Xu. Policy Gradient Converges to the Globally Optimal Policy for Nearly Linear-Quadratic Regulators. Revision, SIAM Control and Optimization, 2024.
  • Short version accepted by NeurIPS Workshop Optimization for Machine Learning, 2022.
• Yinbin Han and Zizhuo Wang. Optimal Switching Policy for Batch Servers. Operations Research Letters, 2023.