Promises and Pitfalls of Threshold-based Auto-labeling

Creating large-scale high-quality labeled datasets is a major bottleneck in supervised machine learning workflows. Threshold-based auto-labeling (TBAL), where validation data obtained from humans is used to find a confidence threshold above which the data is machine-labeled, reduces reliance on manu...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Vishwakarma, Harit, Lin, Heguang, Sala, Frederic, Vinayak, Ramya Korlakai
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 22.02.2024
Subjects
Datasets
Labeling
Machine learning
Shelf life
Online AccessGet full text

Cover

More Information
Summary:Creating large-scale high-quality labeled datasets is a major bottleneck in supervised machine learning workflows. Threshold-based auto-labeling (TBAL), where validation data obtained from humans is used to find a confidence threshold above which the data is machine-labeled, reduces reliance on manual annotation. TBAL is emerging as a widely-used solution in practice. Given the long shelf-life and diverse usage of the resulting datasets, understanding when the data obtained by such auto-labeling systems can be relied on is crucial. This is the first work to analyze TBAL systems and derive sample complexity bounds on the amount of human-labeled validation data required for guaranteeing the quality of machine-labeled data. Our results provide two crucial insights. First, reasonable chunks of unlabeled data can be automatically and accurately labeled by seemingly bad models. Second, a hidden downside of TBAL systems is potentially prohibitive validation data usage. Together, these insights describe the promise and pitfalls of using such systems. We validate our theoretical guarantees with extensive experiments on synthetic and real datasets.
ISSN:2331-8422