Boundary-Guided Black-Box Fairness Testing

Ziqiang Yin, Wentian Zhao*, Tian Song

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Although deep learning models have achieved outstanding performance in many applications, there are still concerns about their fairness. A series of fairness testing methods, which evaluate the fairness of deep learning models by generating discriminatory samples, have been proposed. However, these methods either neglect the naturalness of discriminatory samples or roughly select natural discriminatory samples, leading to a decrease in efficiency. In this paper, we introduce a boundary-guided black-box fairness testing method to effectively generate individual discriminatory samples with high efficiency and enhanced naturalness. Our boundary-guided method involves a global exploration phase, which explores multiple paths from the initial samples to the surrogate decision boundary of the target model, imitated from the semantic latent space of a generative adversarial network (GAN). Then, a local perturbation phase explores the nearby space around a given sample for identifying potential discriminatory samples. Extensive experiments on various datasets demonstrate that our approach outperforms state-of-the-art methods in terms of efficiency and effectiveness while maintaining high naturalness.

Original languageEnglish
Title of host publicationProceedings - 2024 IEEE 48th Annual Computers, Software, and Applications Conference, COMPSAC 2024
EditorsHossain Shahriar, Hiroyuki Ohsaki, Moushumi Sharmin, Dave Towey, AKM Jahangir Alam Majumder, Yoshiaki Hori, Ji-Jiang Yang, Michiharu Takemoto, Nazmus Sakib, Ryohei Banno, Sheikh Iqbal Ahamed
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1230-1239
Number of pages10
ISBN (Electronic)9798350376968
DOIs
Publication statusPublished - 2024
Event48th IEEE Annual Computers, Software, and Applications Conference, COMPSAC 2024 - Osaka, Japan
Duration: 2 Jul 20244 Jul 2024

Publication series

NameProceedings - 2024 IEEE 48th Annual Computers, Software, and Applications Conference, COMPSAC 2024

Conference

Conference48th IEEE Annual Computers, Software, and Applications Conference, COMPSAC 2024
Country/TerritoryJapan
CityOsaka
Period2/07/244/07/24

Keywords

  • Boundary-Guided Method
  • Fairness Testing
  • Individual Discriminatory Samples

Fingerprint

Dive into the research topics of 'Boundary-Guided Black-Box Fairness Testing'. Together they form a unique fingerprint.

Cite this

Yin, Z., Zhao, W., & Song, T. (2024). Boundary-Guided Black-Box Fairness Testing. In H. Shahriar, H. Ohsaki, M. Sharmin, D. Towey, AKM. J. A. Majumder, Y. Hori, J.-J. Yang, M. Takemoto, N. Sakib, R. Banno, & S. I. Ahamed (Eds.), Proceedings - 2024 IEEE 48th Annual Computers, Software, and Applications Conference, COMPSAC 2024 (pp. 1230-1239). (Proceedings - 2024 IEEE 48th Annual Computers, Software, and Applications Conference, COMPSAC 2024). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/COMPSAC61105.2024.00163