On Combining Expert Demonstrations in Imitation Learning via Optimal Transport
This addresses a specific bottleneck in imitation learning for robotics or control tasks, but it is incremental as it builds on existing optimal transport methods.
The paper tackles the problem of combining multiple expert demonstrations in imitation learning, which is typically done by concatenating trajectories but fails with multi-modal data, and proposes a method using multi-marginal optimal transport to provide a geometric average, showing on OpenAI Gym environments that the standard approach is not always optimal.
Imitation learning (IL) seeks to teach agents specific tasks through expert demonstrations. One of the key approaches to IL is to define a distance between agent and expert and to find an agent policy that minimizes that distance. Optimal transport methods have been widely used in imitation learning as they provide ways to measure meaningful distances between agent and expert trajectories. However, the problem of how to optimally combine multiple expert demonstrations has not been widely studied. The standard method is to simply concatenate state (-action) trajectories, which is problematic when trajectories are multi-modal. We propose an alternative method that uses a multi-marginal optimal transport distance and enables the combination of multiple and diverse state-trajectories in the OT sense, providing a more sensible geometric average of the demonstrations. Our approach enables an agent to learn from several experts, and its efficiency is analyzed on OpenAI Gym control environments and demonstrates that the standard method is not always optimal.