Increasing Confidence in Adversarial Robustness Evaluations

Published in NeurIPS 2022, 2022

Zimmermann, R. S., Brendel, W., Tramer, F., and Carlini, N., Increasing Confidence in Adversarial Robustness Evaluations. NeurIPS 2022.

Hundreds of defenses have been proposed to make deep neural networks robust against minimal (adversarial) input perturbations. However, only a handful of these defenses held up their claims because correctly evaluating robustness is extremely challenging: Weak attacks often fail to find adversarial examples even if they unknowingly exist, thereby making a vulnerable network look robust. In this paper, we propose a test to identify weak attacks and, thus, weak defense evaluations. Our test slightly modifies a neural network to guarantee the existence of an adversarial example for every sample. Consequentially, any correct attack must succeed in breaking this modified network. For eleven out of thirteen previously-published defenses, the original evaluation of the defense fails our test, while stronger attacks that break these defenses pass it. We hope that attack unit tests - such as ours - will be a major component in future robustness evaluations and increase confidence in an empirical field that is currently riddled with skepticism.

Project website    Conference paper

@inproceedings{
 zimmermann2022increasing,
 title={   Increasing Confidence in Adversarial
  Robustness Evaluations
 },
 author={   Roland S. Zimmermann and
  Wieland Brendel and
  Florian Tramer and
  Nicholas Carlini
 },
 booktitle={   Thirty-Sixth Conference on Neural
  Information Processing Systems
 },
 year={2022},
 url={https://openreview.net/forum?id=NkK4i91VWp}
}