Deep Learning with Label Differential Privacy

TL;DR

This paper introduces RRWithPrior, an improved randomized response algorithm for label differential privacy in neural networks, achieving better accuracy while maintaining privacy guarantees, supported by empirical and theoretical analysis.

Contribution

The paper proposes RRWithPrior, a novel algorithm that enhances label differential privacy in neural networks, outperforming previous private baselines and providing theoretical insights.

Findings

RRWithPrior improves model accuracy under label privacy.

Using priors further enhances privacy-utility trade-off.

LabelDP is theoretically easier to protect than input privacy.

Abstract

The Randomized Response (RR) algorithm is a classical technique to improve robustness in survey aggregation, and has been widely adopted in applications with differential privacy guarantees. We propose a novel algorithm, Randomized Response with Prior (RRWithPrior), which can provide more accurate results while maintaining the same level of privacy guaranteed by RR. We then apply RRWithPrior to learn neural networks with label differential privacy (LabelDP), and show that when only the label needs to be protected, the model performance can be significantly improved over the previous state-of-the-art private baselines. Moreover, we study different ways to obtain priors, which when used with RRWithPrior can additionally improve the model performance, further reducing the accuracy gap between private and non-private models. We complement the empirical results with theoretical analysis showing that LabelDP is provably easier than protecting both the inputs and labels.