Research talk: Safe reinforcement learning using advantage-based intervention

Microsoft Research - Research talk: Safe reinforcement learning using advantage-based intervention