Fast Policy Learning through Imitation and Reinforcement
This work addresses policy learning efficiency for robotics or AI agents, but it is incremental as it builds on existing RL and IL frameworks.
The paper tackles the problem of combining imitation learning (IL) and reinforcement learning (RL) to overcome suboptimal expert demonstrations, proposing LOKI, a strategy that switches from IL to RL at a randomized time, which learns to outperform suboptimal experts and converges faster than policy gradient alone in simulated environments.
Imitation learning (IL) consists of a set of tools that leverage expert demonstrations to quickly learn policies. However, if the expert is suboptimal, IL can yield policies with inferior performance compared to reinforcement learning (RL). In this paper, we aim to provide an algorithm that combines the best aspects of RL and IL. We accomplish this by formulating several popular RL and IL algorithms in a common mirror descent framework, showing that these algorithms can be viewed as a variation on a single approach. We then propose LOKI, a strategy for policy learning that first performs a small but random number of IL iterations before switching to a policy gradient RL method. We show that if the switching time is properly randomized, LOKI can learn to outperform a suboptimal expert and converge faster than running policy gradient from scratch. Finally, we evaluate the performance of LOKI experimentally in several simulated environments.