AI Lecture Series – Thiago D. Simão

We were pleased to welcome Assistant Professor Thiago D. Simão from Eindhoven University of Technology (TU/e) for a guest lecture on Safety and Reliability in Reinforcement Learning.

Thiago D. Simão (he/him) is an Assistant Professor in the Department of Mathematics and Computer Science at TU/e. He received his Ph.D. from Delft University of Technology and worked as a Postdoctoral Researcher at Radboud University Nijmegen. His research focuses on improving the reliability of AI systems, with particular interest in safe reinforcement learning and ensuring robust performance under uncertainty. In his talk, Prof. Simão explored how to make reinforcement learning (RL) agents more reliable. A key focus was on Safe Policy Improvement, a framework that ensures a reliable performance baseline when learning from offline data. He presented methods that integrate prior knowledge to improve learning efficiency and help maintain safety constraints, especially in offline contexts, where only historical data is available. The talk also addressed challenges such as breaking down complicated tasks and ensuring that a curriculum of simple to more and more complex tasks is learned while minimizing any violations of safety constraints. With a research focus on trustworthy AI, Prof. Simão offered valuable insights into how reinforcement learning can be more robust and applicable to complex, uncertain environments. His presentation sparked thoughtful discussion among attendees and contributed meaningfully to the ongoing academic exchange at RUB.