Generative Adversarial Imitation Learning
This addresses the challenge of imitation learning for AI systems by providing a more efficient alternative to indirect methods, though it is incremental in improving model-free approaches.
The paper tackles the problem of learning a policy from expert behavior without direct interaction or reinforcement signals, proposing a direct framework that yields significant performance gains over existing model-free methods in complex, high-dimensional environments.
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 reinforcement learning, 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 reinforcement learning. 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 imitation learning algorithm that obtains significant performance gains over existing model-free methods in imitating complex behaviors in large, high-dimensional environments.