Single Life Reinforcement Learning - I Hadn't Seen It Before

We want a system to adapt in a new environment at test time, and it has some previous experience. We want the robot to be able to do the task in kind of a single life or in a single episode. The other thing that's different about this is that typically in reinforcement learning, there's this trial and error process. And so the really outcome of that process is this policy that can do the task and can reliably do so.