Active Learning with Task-Driven Representations for Messy Pools

TL;DR

This paper proposes updating representations during active learning to better handle messy data pools, leading to improved performance over fixed, unsupervised representations.

Contribution

It introduces task-driven, periodically updated representations in active learning, enhancing effectiveness in messy data pools compared to traditional fixed representations.

Findings

Task-driven representations outperform fixed unsupervised ones.

Two strategies for updating representations show significant performance gains.

Periodic updates improve active learning efficiency in messy pools.

Abstract

Active learning has the potential to be especially useful for messy, uncurated pools where datapoints vary in relevance to the target task. However, state-of-the-art approaches to this problem currently rely on using fixed, unsupervised representations of the pool, focusing on modifying the acquisition function instead. We show that this model setup can undermine their effectiveness at dealing with messy pools, as such representations can fail to capture important information relevant to the task. To address this, we propose using task-driven representations that are periodically updated during the active learning process using the previously collected labels. We introduce two specific strategies for learning these representations, one based on directly learning semi-supervised representations and the other based on supervised fine-tuning of an initial unsupervised representation. We find that both significantly improve empirical performance over using unsupervised or pretrained representations.