Differentially private correlation clustering

Files
bun21a.pdf(308.29 KB)
Published version
Date
2021-07-18
DOI
Authors
Bun, Mark
Elias, Marek
Kulkarni, Janardhan
Version
OA Version
Citation
M. Bun, M. Elias, J. Kulkarni. 2021. "Differentially private correlation clustering." Proceedings of the 38th International Conference on Machine Learning, PMLR 139:1136-1146, 2021.
Abstract
Correlation clustering is a widely used technique in unsupervised machine learning. Motivated by applications where individual privacy is a concern, we initiate the study of differentially private correlation clustering. We propose an algorithm that achieves subquadratic additive error compared to the optimal cost. In contrast, straightforward adaptations of existing non-private algorithms all lead to a trivial quadratic error. Finally, we give a lower bound showing that any pure differentially private algorithm for correlation clustering requires additive error of Ω (n).
Description
License
Copyright 2021 by the author(s).