Imitation Learning

Learning action mapping directly from demonstrations.

Inferring reward function from observed behavior.

Directly optimizing control policies.

Optimizing continuous action sequences.

Modifying reward to accelerate learning.

Reward only given upon task completion.