arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2203.11396cs.CLcs.AIcs.LG
收藏

一种无需域内标记的文本域外检测方法

Towards Textual Out-of-Domain Detection without In-Domain Labels

Amazon Alexa AI(亚马逊Alexa人工智能部门)

Di Jin, Shuyang Gao, Seokhwan Kim, Yang Liu, Dilek Hakkani-Tur

摘要:

在许多现实环境中,机器学习模型需要识别域外(OOD)的用户输入,以避免执行错误的操作。这项工作的重点是OOD检测的一个具有挑战性的情况下,没有标签的域内数据是可访问的(例如,没有意图分类任务的意图标签)。为此,我们首先评估不同的语言模型为基础的方法,预测一个序列的令牌的可能性。此外,我们提出了一种新的表示学习为基础的方法,结合无监督聚类和对比学习,使更好的数据表示OOD检测可以学习。通过大量的实验,我们证明了这种方法可以显着优于基于似然的方法,甚至可以竞争的最先进的监督方法与标签信息。

英文摘要:

In many real-world settings, machine learning models need to identify user inputs that are out-of-domain (OOD) so as to avoid performing wrong actions. This work focuses on a challenging case of OOD detection, where no labels for in-domain data are accessible (e.g., no intent labels for the intent classification task). To this end, we first evaluate different language model based approaches that predict likelihood for a sequence of tokens. Furthermore, we propose a novel representation learning based method by combining unsupervised clustering and contrastive learning so that better data representations for OOD detection can be learned. Through extensive experiments, we demonstrate that this method can significantly outperform likelihood-based methods and can be even competitive to the state-of-the-art supervised approaches with label information.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑