Active Fine-Tuning of Multi-Task Policies

TL;DR

This paper introduces AMF, an active learning algorithm that adaptively selects demonstrations to efficiently fine-tune multi-task policies, improving performance with limited data in complex environments.

Contribution

The paper proposes AMF, a novel active learning approach for multi-task policy fine-tuning that maximizes information gain and provides theoretical performance guarantees.

Findings

AMF outperforms baseline methods in efficiency and accuracy.

AMF effectively adapts to complex, high-dimensional environments.

Theoretical guarantees support the effectiveness of AMF.

Abstract

Pre-trained generalist policies are rapidly gaining relevance in robot learning due to their promise of fast adaptation to novel, in-domain tasks. This adaptation often relies on collecting new demonstrations for a specific task of interest and applying imitation learning algorithms, such as behavioral cloning. However, as soon as several tasks need to be learned, we must decide which tasks should be demonstrated and how often? We study this multi-task problem and explore an interactive framework in which the agent adaptively selects the tasks to be demonstrated. We propose AMF (Active Multi-task Fine-tuning), an algorithm to maximize multi-task policy performance under a limited demonstration budget by collecting demonstrations yielding the largest information gain on the expert policy. We derive performance guarantees for AMF under regularity assumptions and demonstrate its empirical effectiveness to efficiently fine-tune neural policies in complex and high-dimensional environments.