PAC-learning in the presence of evasion adversaries

TL;DR

This paper extends the PAC-learning framework to include evasion adversaries, analyzing the impact on learnability and VC-dimension, and providing bounds on sample complexity in adversarial settings.

Contribution

It introduces the adversarial VC-dimension, derives its value for halfspace classifiers, and explores how adversaries affect the learnability of hypothesis classes.

Findings

Adversarial VC-dimension can match or differ from standard VC-dimension.

Sample complexity bounds extend to adversarial settings using the adversarial VC-dimension.

For halfspace classifiers, the adversarial VC-dimension equals the standard VC-dimension.

Abstract

The existence of evasion attacks during the test phase of machine learning algorithms represents a significant challenge to both their deployment and understanding. These attacks can be carried out by adding imperceptible perturbations to inputs to generate adversarial examples and finding effective defenses and detectors has proven to be difficult. In this paper, we step away from the attack-defense arms race and seek to understand the limits of what can be learned in the presence of an evasion adversary. In particular, we extend the Probably Approximately Correct (PAC)-learning framework to account for the presence of an adversary. We first define corrupted hypothesis classes which arise from standard binary hypothesis classes in the presence of an evasion adversary and derive the Vapnik-Chervonenkis (VC)-dimension for these, denoted as the adversarial VC-dimension. We then show that sample complexity upper bounds from the Fundamental Theorem of Statistical learning can be extended to the case of evasion adversaries, where the sample complexity is controlled by the adversarial VC-dimension. We then explicitly derive the adversarial VC-dimension for halfspace classifiers in the presence of a sample-wise norm-constrained adversary of the type commonly studied for evasion attacks and show that it is the same as the standard VC-dimension, closing an open question. Finally, we prove that the adversarial VC-dimension can be either larger or smaller than the standard VC-dimension depending on the hypothesis class and adversary, making it an interesting object of study in its own right.