Reinforcement Learning. Decision-Making & Optimal Control.
Research at the intersection of reinforcement learning and engineering optimization.
About
Bingyang Yan received his PhD from China University of Geosciences in June 2026 under the supervision of Prof. Zhi Zhong. From Sep 2024 to Feb 2026, he was a visiting PhD student at the University of Alberta, working with Prof. Jeff Boisvert.
Research
His research focused on model-free reinforcement learning for optimal control in complex engineering systems, particularly those involving high-dimensional spatiotemporal dynamics and computationally expensive simulations. His expertise lies in modeling these dynamics and developing learned surrogate environments to accelerate policy learning. By combining physics-based simulation, surrogate modeling, and policy optimization, he addressed computational bottlenecks in continuous control and engineering decision-making. He implemented these methods primarily in Python and PyTorch.
Interests
- Multi-Agent RL, Hierarchical RL, Safe RL, Model-Based RL (World Models, MPC)
- RL with GNNs and Transformers
- Agentic RL, LLM (Post-Training), VLM
- Intelligent Decision-Making and Optimal Control in Engineering and Industrial Systems