Creating Advice-Taking Reinforcement Learners

Abstract

Learning from reinforcements is a promising approach for creating intelligent agents. However, reinforcement learning usually requires a large number of training episodes. We present …