Inference and Uncertainty Quantification for Offline Reinforcement Learning

Reinforcement learning (RL) agents for sequential decision-making in finite-state systems. For
real-word deployment it is necessary to quantify uncertainty in the outcomes. We address quantifying
epistemic as well as aleatoric uncertainty in finite-state environments with limited data (offline RL).
Apply methods to interpretable gridworlds and data for clinical decision support systems

Brain behaviour Lab
AI machine learning