:::
AIoTC

Instance-Optimal Policy Testing in Reinforcement Learning


  • 講者 : 王柏安 教授
  • 日期 : 2026/09/01 (Tue.) 10:00~11:30
  • 地點 : 資創中心122 演講廳
  • 邀請人 : 逄愛君主任
Abstract

Policy testing asks whether the value of a given policy exceeds a prescribed threshold while using as few observations as possible. Although existing methods provide worst-case guarantees, developing an instance-optimal and computationally tractable algorithm is challenging because the corresponding lower-bound optimization problem is non-convex.

In this talk, I first study single-policy testing in discounted Markov decision processes under a generative model. I derive an instance-specific sample-complexity lower bound and show that its seemingly difficult optimization problem can be reformulated using a reversed Markov decision process. This reformulation exchanges the objective and constraint, allowing us to apply policy-gradient methods. Building on this idea, I introduce Policy Testing via Simulation and Tracking (PTST), an algorithm that adaptively selects informative state–action pairs, tracks the optimal sampling allocation, and stops once sufficient evidence has been collected. PTST is computationally tractable and asymptotically matches the instance-specific lower bound.

Finally, I discuss joint policy testing in the forward model, where the learner must evaluate several policies using a single adaptive trajectory. I present PT-ACE, which addresses the additional navigation and flow constraints and achieves instance-specific asymptotic optimality in this more challenging setting.

Bio
Dr. Po-An Wang is an assistant professor affiliated with the Institute of Statistics and Data Science at National Tsing Hua University. Before joining this institution, He was a PhD student at KTH Royal Institute of Technology, where he worked under the guidance of Prof. Alexandre Proutiere. In June 2023, he embarked on an enriching internship journey with CyberAgent in Japan. Prior to his academic pursuits at KTH, he gained valuable research experience as an assistant in Chi-Jen Lu’s lab at Academia Sinica.