Inverse Reinforcement Learning

Modifying reward to accelerate learning.

Reward only given upon task completion.

Learning policies from expert demonstrations.

Learning action mapping directly from demonstrations.