Besting the Black-Box: Barrier Zones for Adversarial Example Defense

Kaleel Mahmood, Phuong Ha Nguyen, Lam M. Nguyen, Thanh Nguyen, Marten van Dijk

Abstract

Adversarial machine learning defenses have primarily been focused on mitigating static, white-box attacks. However, it remains an open question whether such defenses are robust under an adaptive black-box adversary. In this paper, we specifically focus on the black-box threat model and make the following contributions: First we develop an enhanced adaptive black-box attack which is experimentally shown to be <inline-formula> <tex-math notation="LaTeX">$\geq 30\%$ </tex-math></inline-formula> more effective than the original adaptive black-box attack proposed by Papernot <i>et al.</i> For our second contribution, we test 10 recent defenses using our new attack and propose our own black-box defense (barrier zones). We show that our defense based on barrier zones offers significant improvements in security over state-of-the-art defenses. This improvement includes greater than 85&#x0025; robust accuracy against black-box boundary attacks, transfer attacks and our new adaptive black-box attack, for the datasets we study. For completeness, we verify our claims through extensive experimentation with 10 other defenses using three adversarial models (14 different black-box attacks) on two datasets (CIFAR-10 and Fashion-MNIST).
BibTeX
@article{mahmood2022besting,
  author = {Kaleel Mahmood and Phuong Ha Nguyen and Lam M. Nguyen and Thanh Nguyen and Marten van Dijk},
  title = {{Besting the Black-Box: Barrier Zones for Adversarial Example Defense}},
  journal = {IEEE Access},
  year = {2022},
  doi = {10.1109/access.2021.3138966},
}

Loading related content...

Figure