Motivating the Rules of the Game for Adversarial Example Research

TL;DR

This paper critiques current adversarial example research for lacking realistic threat models and proposes a taxonomy of attacker motivations and constraints to guide more meaningful future evaluations.

Contribution

It introduces a taxonomy of attacker motivations, constraints, and abilities, and offers recommendations for clearer threat models and evaluation standards in adversarial research.

Findings

Current defenses are based on toy models that lack real-world relevance.

A taxonomy of attacker capabilities helps clarify security threats.

Recommendations for standardized evaluation practices.

Abstract

Advances in machine learning have led to broad deployment of systems with impressive performance on important problems. Nonetheless, these systems can be induced to make errors on data that are surprisingly similar to examples the learned system handles correctly. The existence of these errors raises a variety of questions about out-of-sample generalization and whether bad actors might use such examples to abuse deployed systems. As a result of these security concerns, there has been a flurry of recent papers proposing algorithms to defend against such malicious perturbations of correctly handled examples. It is unclear how such misclassifications represent a different kind of security problem than other errors, or even other attacker-produced examples that have no specific relationship to an uncorrupted input. In this paper, we argue that adversarial example defense papers have, to date, mostly considered abstract, toy games that do not relate to any specific security concern. Furthermore, defense papers have not yet precisely described all the abilities and limitations of attackers that would be relevant in practical security. Towards this end, we establish a taxonomy of motivations, constraints, and abilities for more plausible adversaries. Finally, we provide a series of recommendations outlining a path forward for future work to more clearly articulate the threat model and perform more meaningful evaluation.