Augment then Smooth: Reconciling Differential Privacy with Certified Robustness

TL;DR

This paper introduces DP-CERT, a simple method that combines differential privacy and certified robustness by integrating randomized smoothing into private training, improving certified accuracy and reducing Lipschitz constants.

Contribution

DP-CERT is a novel, straightforward approach that effectively merges differential privacy with certified robustness, outperforming prior complex training schemes.

Findings

Up to 2.5% increase in certified accuracy on CIFAR10

Larger certifiable radii correlate with smaller local Lipschitz constants

DP-CERT reduces Lipschitz constants compared to other private training methods

Abstract

Machine learning models are susceptible to a variety of attacks that can erode trust, including attacks against the privacy of training data, and adversarial examples that jeopardize model accuracy. Differential privacy and certified robustness are effective frameworks for combating these two threats respectively, as they each provide future-proof guarantees. However, we show that standard differentially private model training is insufficient for providing strong certified robustness guarantees. Indeed, combining differential privacy and certified robustness in a single system is non-trivial, leading previous works to introduce complex training schemes that lack flexibility. In this work, we present DP-CERT, a simple and effective method that achieves both privacy and robustness guarantees simultaneously by integrating randomized smoothing into standard differentially private model training. Compared to the leading prior work, DP-CERT gives up to a 2.5% increase in certified accuracy for the same differential privacy guarantee on CIFAR10. Through in-depth per-sample metric analysis, we find that larger certifiable radii correlate with smaller local Lipschitz constants, and show that DP-CERT effectively reduces Lipschitz constants compared to other differentially private training methods. The code is available at github.com/layer6ai-labs/dp-cert.