Active Imitation Learning via Reduction to I.I.D. Active Learning

TL;DR

This paper presents a new active imitation learning method that reduces the problem to i.i.d. active learning, significantly decreasing expert queries needed for policy learning and demonstrating effectiveness across multiple domains.

Contribution

It introduces a reduction-based approach to active imitation learning leveraging i.i.d. active learning techniques, with theoretical analysis and a practical algorithm.

Findings

Reduced label complexity compared to passive learning

Effective performance in four test domains

Theoretical bounds for non-stationary and stationary policies

Abstract

In standard passive imitation learning, the goal is to learn a target policy by passively observing full execution trajectories of it. Unfortunately, generating such trajectories can require substantial expert effort and be impractical in some cases. In this paper, we consider active imitation learning with the goal of reducing this effort by querying the expert about the desired action at individual states, which are selected based on answers to past queries and the learner's interactions with an environment simulator. We introduce a new approach based on reducing active imitation learning to i.i.d. active learning, which can leverage progress in the i.i.d. setting. Our first contribution, is to analyze reductions for both non-stationary and stationary policies, showing that the label complexity (number of queries) of active imitation learning can be substantially less than passive learning. Our second contribution, is to introduce a practical algorithm inspired by the reductions, which is shown to be highly effective in four test domains compared to a number of alternatives.