Active Sentence Learning by Adversarial Uncertainty Sampling in Discrete Space

TL;DR

This paper introduces AUSDS, an efficient active learning method that uses adversarial attacks in latent space to select informative sentences for annotation, significantly speeding up the process while improving effectiveness.

Contribution

The paper proposes AUSDS, a novel active learning approach that leverages adversarial attacks in latent space for faster and more effective sample selection in sentence understanding.

Findings

AUSDS achieves over 10x speedup compared to traditional methods.

AUSDS outperforms baseline methods on five datasets.

The approach effectively identifies informative samples for annotation.

Abstract

Active learning for sentence understanding aims at discovering informative unlabeled data for annotation and therefore reducing the demand for labeled data. We argue that the typical uncertainty sampling method for active learning is time-consuming and can hardly work in real-time, which may lead to ineffective sample selection. We propose adversarial uncertainty sampling in discrete space (AUSDS) to retrieve informative unlabeled samples more efficiently. AUSDS maps sentences into latent space generated by the popular pre-trained language models, and discover informative unlabeled text samples for annotation via adversarial attack. The proposed approach is extremely efficient compared with traditional uncertainty sampling with more than 10x speedup. Experimental results on five datasets show that AUSDS outperforms strong baselines on effectiveness.