Breaking Fair Binary Classification with Optimal Flipping Attacks

Minimizing risk with fairness constraints is one of the popular approaches to learning a fair classifier. Recent works showed that this approach yields an unfair classifier if the training set is corrupted. In this work, we study the minimum amount of data corruption required for a successful flippi...

Full description

Saved in:
Bibliographic Details
Published in2022 IEEE International Symposium on Information Theory (ISIT) pp. 1453 - 1458
Main Authors Jo, Changhun, Sohn, Jy-Yong, Lee, Kangwook
Format Conference Proceeding
LanguageEnglish
Published IEEE 26.06.2022
Subjects
Classification algorithms
Collaborative work
Computational efficiency
Computational modeling
Perturbation methods
Training
Upper bound
Online AccessGet full text

Cover

More Information
Summary:Minimizing risk with fairness constraints is one of the popular approaches to learning a fair classifier. Recent works showed that this approach yields an unfair classifier if the training set is corrupted. In this work, we study the minimum amount of data corruption required for a successful flipping attack. First, we find lower/upper bounds on this quantity and show that these bounds are tight when the target model is the unique unconstrained risk minimizer. Second, we propose a computationally efficient data poisoning attack algorithm that can compromise the performance of fair learning algorithms.
ISSN:2157-8117
DOI:10.1109/ISIT50566.2022.9834475