Yuexin Bian (边玥心)
I am a PhD student in the Department of Electrical and Computer Engineering at UC San Diego, where I work under the guidance of Prof. Yuanyuan Shi. I received my B.S. in Electrical Engineering from Zhejiang University, China.
My research lies at the intersection of optimization and reinforcement learning, with a focus on developing model-guided learning methods that bridge the gap between theoretical foundations and practical applications. I am particularly interested in creating intelligent systems that can make efficient, reliable, and interpretable decisions in complex real-world scenarios, especially in energy systems and beyond.
Research
My research focuses on developing model-guided learning methods that combine the power of reinforcement learning with domain knowledge from optimization and control theory.
I am particularly passionate about creating algorithms that integrate principled mathematical structures and problem-specific constraints into the learning process. The goal is to develop methods that are not only sample-efficient but also capable of making intelligent, reliable, and interpretable decisions in complex real-world systems.
News
- Nov 2025: 🎉 Our paper “DiffOP: Reinforcement Learning of Optimization-Based Control Policies via Implicit Policy Gradients” has been accepted to AAAI 2026.
- Nov 2025: 📢 I will be presenting at INFORMS 2026.
- Nov 2025: 🎉 Our paper “Data-driven operator learning for energy-efficient building control” has been accepted to Applied Energy.
- Sep 2025: 🎉 Our paper “Revisiting Reinforcement Learning for LLM Reasoning from A Cross-Domain Perspective” has been accepted to NeurIPS 2025.
Experience
-
Applied Scientist Intern | Amazon Supply Chain Optimization Team (SCOT)
Jun 2025 – Sep 2025, Seattle
Working on inventory optimization with visual fullness. -
Applied Scientist Intern | Amazon Grocery Optimization Team
Jun 2024 – Sep 2024, Seattle
Developed robust macroplanogram design solutions to optimize store layouts and improve customer experience.