Efficient Robust Conformal Prediction via Lipschitz-Bounded Networks

TL;DR

This paper introduces a novel Lipschitz-bounded network approach to improve the efficiency and size of robust conformal prediction sets under adversarial attacks, achieving state-of-the-art results on large-scale datasets like ImageNet.

Contribution

The paper proposes a new Lipschitz-bounded network method for robust conformal prediction, enhancing efficiency and set size reduction in large-scale adversarial scenarios.

Findings

Outperforms existing methods in set size and computational efficiency on ImageNet.

Provides new worst-case coverage bounds for vanilla conformal prediction under attack.

Achieves robustness guarantees with computational efficiency comparable to vanilla CP.

Abstract

Conformal Prediction (CP) has proven to be an effective post-hoc method for improving the trustworthiness of neural networks by providing prediction sets with finite-sample guarantees. However, under adversarial attacks, classical conformal guarantees do not hold anymore: this problem is addressed in the field of Robust Conformal Prediction. Several methods have been proposed to provide robust CP sets with guarantees under adversarial perturbations, but, for large scale problems, these sets are either too large or the methods are too computationally demanding to be deployed in real life scenarios. In this work, we propose a new method that leverages Lipschitz-bounded networks to precisely and efficiently estimate robust CP sets. When combined with a 1-Lipschitz robust network, we demonstrate that our lip-rcp method outperforms state-of-the-art results in both the size of the robust CP sets and computational efficiency in medium and large-scale scenarios such as ImageNet. Taking a different angle, we also study vanilla CP under attack, and derive new worst-case coverage bounds of vanilla CP sets, which are valid simultaneously for all adversarial attack levels. Our lip-rcp method makes this second approach as efficient as vanilla CP while also allowing robustness guarantees.