Back to Search Start Over

Randomness in ML Defenses Helps Persistent Attackers and Hinders Evaluators

Authors :
Lucas, Keane
Jagielski, Matthew
Tramèr, Florian
Bauer, Lujo
Carlini, Nicholas
Publication Year :
2023

Abstract

It is becoming increasingly imperative to design robust ML defenses. However, recent work has found that many defenses that initially resist state-of-the-art attacks can be broken by an adaptive adversary. In this work we take steps to simplify the design of defenses and argue that white-box defenses should eschew randomness when possible. We begin by illustrating a new issue with the deployment of randomized defenses that reduces their security compared to their deterministic counterparts. We then provide evidence that making defenses deterministic simplifies robustness evaluation, without reducing the effectiveness of a truly robust defense. Finally, we introduce a new defense evaluation framework that leverages a defense's deterministic nature to better evaluate its adversarial robustness.

Details

Database :
arXiv
Publication Type :
Report
Accession number :
edsarx.2302.13464
Document Type :
Working Paper