Natural Language Understanding Wiki
Advertisement

In general reinforcement learning, imitation learning makes use of (possibly suboptimal) referent policy to train an agent, as opposed to starting from scratch.

In NLP, imitation learning is usually associated with learning to search.