Enhancing Adversarial Robustness via Test-time Transformation Ensembling

TL;DR

This paper demonstrates that Test-time Transformation Ensembling (TTE) significantly enhances adversarial robustness of deep learning models against various attacks without sacrificing accuracy on clean data.

Contribution

The study provides a comprehensive empirical analysis showing TTE's effectiveness in improving adversarial robustness and certified robustness without re-training models.

Findings

TTE improves robustness against multiple attacks.

TTE maintains accuracy on clean samples.

TTE benefits transfer to certified robustness domain.

Abstract

Deep learning models are prone to being fooled by imperceptible perturbations known as adversarial attacks. In this work, we study how equipping models with Test-time Transformation Ensembling (TTE) can work as a reliable defense against such attacks. While transforming the input data, both at train and test times, is known to enhance model performance, its effects on adversarial robustness have not been studied. Here, we present a comprehensive empirical study of the impact of TTE, in the form of widely-used image transforms, on adversarial robustness. We show that TTE consistently improves model robustness against a variety of powerful attacks without any need for re-training, and that this improvement comes at virtually no trade-off with accuracy on clean samples. Finally, we show that the benefits of TTE transfer even to the certified robustness domain, in which TTE provides sizable and consistent improvements.