1 paper touches this idea.
Papers
Teaching robots by thumbs-up: a smarter way to pick what to try next
Related concepts
Ideas that show up alongside Model-Based Reinforcement Learning.