2019 10-11 17:25

To be precise, the “imitation learning” is the general problem of learning from expert demonstration (LfD). There are 2 names derived from such a description, which are Imitation Learning and Apprenticeship Learning due to historical reasons. Usually, apprenticeship learning is mentioned in the context of “Apprenticeship learning via inverse reinforcement learning (IRL)” which recovers the reward function and learns policies from it, while imitation learning began with behavior cloning that learn the policy directly (ref). However, with the development of related researches, “imitation learning” is always used to represent the general LfD problem setting, which is also our view of point.