Min Wen

Papers

Total Citations

H-Index

About

Min Wen is a researcher specializing in the intersection of formal methods, reinforcement learning, and control synthesis, with a particular focus on developing principled approaches to autonomous decision-making under complex constraints. His most recognized contribution, "Correct-by-synthesis Reinforcement Learning with Temporal Logic Constraints" (2015), has accumulated 54 citations and represents a foundational effort in bridging the gap between classical formal verification and modern machine learning. In this work, Wen and collaborators tackled the challenge of synthesizing reactive controllers that optimize unknown performance criteria while provably satisfying temporal logic specifications — a critical concern in safety-critical systems such as robotics and autonomous control. By decomposing the problem into tractable sub-problems, his framework offers correctness guarantees that traditional reward-based reinforcement learning approaches cannot provide. His subsequent work, "Reinforcement Learning With High-Level Task Specifications" (2019), extends these ideas by addressing fundamental shortcomings of reward-centric RL, advocating instead for task-level guarantees. Collectively, Wen's research contributes meaningfully to the growing field of safe and verifiable reinforcement learning, offering tools that are increasingly relevant to practitioners building reliable autonomous systems.