Machine Unlearning with Minimal Gradient Dependence for High Unlearning Ratios
Communications of the ACM · 01.07.2026 · 69 (7) · 74-81
DOI: 10.1145/3793867 ↗Kurz gefasst
Fachartikel in der Zeitschrift Communications of the ACM, online erschienen am 30.06.2026. Verfasst von Tao Huang und 6 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
In the context of machine unlearning, the primary challenge lies in effectively removing traces of private data from trained models while maintaining model performance and security against privacy attacks like membership inference attacks. Traditional gradient-based unlearning methods often rely on extensive historical gradients, which becomes impractical with high unlearning ratios and may reduce the effectiveness of computation. To address these limitations, we introduce Mini-Unlearning, based on our theoretical analysis. Our analysis suggests that unlearned parameters correlate with retrained parameters through contraction mapping, a finding that guides our algorithm design. Our method, Mini-Unlearning, utilizes a minimal subset of historical gradients and leverages this contraction mapping to facilitate scalable, efficient unlearning. This lightweight, scalable method significantly enhances model accuracy and strengthens resistance to membership inference attacks. Our experiments demonstrate that Mini-Unlearning works well under higher unlearning ratios and outperforms existing techniques in accuracy and security, offering a promising solution for applications requiring robust unlearning capabilities.
Zitieren
APA
Huang, T., Chen, Z., Meng, J., Yang, X., Zheng, G., Yi, X., & Khalil, I. (2026). Machine Unlearning with Minimal Gradient Dependence for High Unlearning Ratios. Communications of the ACM, 69(7), 74-81. https://doi.org/10.1145/3793867
Vancouver
Huang T, Chen Z, Meng J, Yang X, Zheng G, Yi X, et al. Machine Unlearning with Minimal Gradient Dependence for High Unlearning Ratios. Communications of the ACM. 2026;69(7):74-81. doi:10.1145/3793867
BibTeX
@article{Huang2026,
author = {Huang, Tao and Chen, Ziyang and Meng, Jiayang and Yang, Xu and Zheng, Guolong and Yi, Xun and Khalil, Ibrahim},
title = {Machine Unlearning with Minimal Gradient Dependence for High Unlearning Ratios},
journal = {Communications of the ACM},
year = {2026},
volume = {69},
number = {7},
pages = {74--81},
doi = {10.1145/3793867}
}
EUVIMED