Adversaries With Incentives: A Strategic Alternative to Adversarial Robustness

TL;DR

This paper introduces a strategic training approach that models adversaries with incentives rather than purely malicious intent, leveraging strategic reasoning to improve robustness and reduce conservativeness in adversarial training.

Contribution

It proposes a novel framework that incorporates opponent incentives into training, enabling more flexible and potentially more effective defenses against a range of adversaries.

Findings

Knowledge of opponent incentives improves robustness.

Potential gains depend on incentive alignment with the task.

Strategic training reduces conservativeness compared to traditional adversarial methods.

Abstract

Adversarial training aims to defend against adversaries: malicious opponents whose sole aim is to harm predictive performance in any way possible. This presents a rather harsh perspective, which we assert results in unnecessarily conservative training. As an alternative, we propose to model opponents as simply pursuing their own goals--rather than working directly against the classifier. Employing tools from strategic modeling, our approach enables knowledge or beliefs regarding the opponent's possible incentives to be used as inductive bias for learning. Accordingly, our method of strategic training is designed to defend against all opponents within an 'incentive uncertainty set'. This resorts to adversarial learning when the set is maximal, but offers potential gains when the set can be appropriately reduced. We conduct a series of experiments that show how even mild knowledge regarding the opponent's incentives can be useful, and that the degree of potential gains depends on how these incentives relate to the structure of the learning task.