Learning a Policy for Opportunistic Active Learning

TL;DR

This paper introduces a reinforcement learning approach to opportunistic active learning for interactive object retrieval, optimizing the trade-off between task success and model improvement for future interactions.

Contribution

It presents a novel reinforcement learning-based policy for opportunistic active learning in interactive tasks, enhancing object retrieval performance.

Findings

Improved object retrieval accuracy using learned policies

Effective balancing of task completion and model learning

Demonstrated benefits over non-opportunistic methods

Abstract

Active learning identifies data points to label that are expected to be the most useful in improving a supervised model. Opportunistic active learning incorporates active learning into interactive tasks that constrain possible queries during interactions. Prior work has shown that opportunistic active learning can be used to improve grounding of natural language descriptions in an interactive object retrieval task. In this work, we use reinforcement learning for such an object retrieval task, to learn a policy that effectively trades off task completion with model improvement that would benefit future tasks.