Towards Bridging the gap between Empirical and Certified Robustness against Adversarial Examples

TL;DR

This paper introduces a novel method that transforms adversarially trained models into certified robust classifiers during inference, effectively combining empirical and certified robustness without sacrificing accuracy.

Contribution

The paper proposes 'Certification through Adaptation' and 'Auto-Noise' techniques to enable certified robustness in adversarially trained models during inference.

Findings

Achieves up to 1.102 and 1.148 average certified radius on CIFAR-10 and ImageNet.

Maintains empirical robustness and accuracy while providing certified guarantees.

Bridges the gap between empirical and certified robustness in adversarial defenses.

Abstract

The current state-of-the-art defense methods against adversarial examples typically focus on improving either empirical or certified robustness. Among them, adversarially trained (AT) models produce empirical state-of-the-art defense against adversarial examples without providing any robustness guarantees for large classifiers or higher-dimensional inputs. In contrast, existing randomized smoothing based models achieve state-of-the-art certified robustness while significantly degrading the empirical robustness against adversarial examples. In this paper, we propose a novel method, called \emph{Certification through Adaptation}, that transforms an AT model into a randomized smoothing classifier during inference to provide certified robustness for $\ell_2$ norm without affecting their empirical robustness against adversarial attacks. We also propose \emph{Auto-Noise} technique that efficiently approximates the appropriate noise levels to flexibly certify the test examples using randomized smoothing technique. Our proposed \emph{Certification through Adaptation} with \emph{Auto-Noise} technique achieves an \textit{average certified radius (ACR) scores} up to $1.102$ and $1.148$ respectively for CIFAR-10 and ImageNet datasets using AT models without affecting their empirical robustness or benign accuracy. Therefore, our paper is a step towards bridging the gap between the empirical and certified robustness against adversarial examples by achieving both using the same classifier.