Adaptive Randomized Smoothing: Certified Adversarial Robustness for Multi-Step Defences

Saiyue Lyu, Shadab Shaikh, Frederick Shpilevskiy, Evan Shelhamer, Mathias Lécuyer

Keywords: Robustness, Adversarial examples, Adaptive defenses, Certified test-time defenses, Randomized Smoothing

TL;DR: Adaptive Randomized Smoothing soundly and flexibly certifies the predictions of test-time adaptive models against adversarial examples while improving certified and standard accuracies.

Abstract:

We propose Adaptive Randomized Smoothing (ARS) to certify the predictions of our test-time adaptive models against adversarial examples.
ARS extends the analysis of randomized smoothing using $f$-Differential Privacy to certify the adaptive composition of multiple steps.
For the first time, our theory covers the sound adaptive composition of general and high-dimensional functions of noisy inputs.
We instantiate ARS on deep image classification to certify predictions against adversarial examples of bounded $L_{\infty}$ norm.
In the $L_{\infty}$ threat model, ARS enables flexible adaptation through high-dimensional input-dependent masking.
We design adaptivity benchmarks, based on CIFAR-10 and CelebA, and show that ARS improves standard test accuracy by 1 to 15\% points.
On ImageNet, ARS improves certified test accuracy by up to 1.6% points over standard RS without adaptivity. Our code is available at [https://github.com/ubc-systopia/adaptive-randomized-smoothing](https://github.com/ubc-systopia/adaptive-randomized-smoothing).

Primary Area: Privacy

Submission Number: 13697