Generative Adversarial Imitation Learning

Consider learning a policy from example expert behavior, without interaction with the expert or access to reinforcement signal. One approach is to recover the expert's cost function with inverse <PRE_TAG>reinforcement learning</POST_TAG>, then extract a policy from that cost function with reinforcement learning. This approach is indirect and can be slow. We propose a new general framework for directly extracting a policy from data, as if it were obtained by reinforcement learning following inverse <PRE_TAG>reinforcement learning</POST_TAG>. We show that a certain instantiation of our framework draws an analogy between imitation learning and generative adversarial networks, from which we derive a model-free <PRE_TAG>imitation learning</POST_TAG> algorithm that obtains significant performance gains over existing model-free methods in imitating complex behaviors in large, high-dimensional environments.