Reinforcement Learning under Unmeasured Confounding In practical reinforcement learning (RL), a representation of the full state which makes the system Markovian and therefore amenable to most existing RL algorithms is not known a priori. Decision makers are often facing so-called … Read more