Differential Privacy has Bounded Impact on Fairness in Classification

2022/10/28 by Mangold, Paul, Perrot, Michaël, Bellet, Aurélien +1 · 4 citations
#Cryptography and Security (cs.CR) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · doi:10.48550/arxiv.2210.16242

Abstract

We theoretically study the impact of differential privacy on fairness in classification. We prove that, given a class of models, popular group fairness measures are pointwise Lipschitz-continuous with respect to the parameters of the model. This result is a consequence of a more general statement on accuracy conditioned on an arbitrary event (such as membership to a sensitive group), which may be of independent interest. We use this Lipschitz property to prove a non-asymptotic bound showing that, as the number of samples increases, the fairness level of private models gets closer to the one of their non-private counterparts. This bound also highlights the importance of the confidence margin of a model on the disparate impact of differential privacy.

Cited by

• Fairness Meets Privacy: Integrating Differential Privacy and Demographic Parity in Multi-class Classification
• A comprehensive review of privacy challenges and anonymization strategies in online social networks
• The Fair Game: Auditing & Debiasing AI Algorithms Over Time
• Private Rate-Constrained Optimization with Applications to Fair Learning

Related

• The Impact of Differential Privacy on Group Disparity Mitigation
• On the Impact of Output Perturbation on Fairness in Binary Linear Classification
• An Empirical Analysis of Fairness Notions under Differential Privacy
• Differentially Private Fair Learning
• A Statistical Viewpoint on Differential Privacy: Hypothesis Testing, Representation and Blackwell's Theorem