(Local) Differential Privacy has NO Disparate Impact on Fairness

Abstract

In recent years, Local Differential Privacy (LDP), a robust privacy-preserving methodology, has gained widespread adoption in real-world applications. With LDP, users can perturb their data on their devices before sending it out for analysis. However, as the collection of multiple sensitive information becomes more prevalent across various industries, collecting a single sensitive attribute under LDP may not be sufficient. Correlated attributes in the data may still lead to inferences about the sensitive attribute. This paper empirically studies the impact of collecting multiple sensitive attributes under LDP on fairness. We propose a novel privacy budget allocation scheme that considers the varying domain size of sensitive attributes. This generally led to a better privacy-utility-fairness trade-off in our experiments than the state-of-art solution. Our results show that LDP leads to slightly improved fairness in learning problems without significantly affecting the performance of the models. We conduct extensive experiments evaluating three benchmark datasets using several group fairness metrics and seven state-of-the-art LDP protocols. Overall, this study challenges the common belief that differential privacy necessarily leads to worsened fairness in machine learning.

Best paper award at DBSec'23. Version of record at https://doi.org/10.1007/978-3-031-37586-6_1

References in corpus (8)

• The NumPy array: a structure for efficient numerical computation
• RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response
• Local, Private, Efficient Protocols for Succinct Histograms
• Machine learning fairness notions: Bridging the gap with real-world applications
• Fairness Score and Process Standardization: Framework for Fairness Certification in Artificial Intelligence Systems
• Differential Privacy and Fairness in Decisions and Learning Tasks: A Survey
• Random Sampling Plus Fake Data: Multidimensional Frequency Estimates With Local Differential Privacy
• (Local) Differential Privacy has NO Disparate Impact on Fairness

Cited by in corpus (1)

• (Local) Differential Privacy has NO Disparate Impact on Fairness