Transferring Adversarial Robustness Through Robust Representation Matching

Authors: 

Pratik Vaishnavi, Stony Brook University; Kevin Eykholt, IBM; Amir Rahmati, Stony Brook University

Abstract: 

With the widespread use of machine learning, concerns over its security and reliability have become prevalent. As such, many have developed defenses to harden neural networks against adversarial examples, imperceptibly perturbed inputs that are reliably misclassified. Adversarial training in which adversarial examples are generated and used during training is one of the few known defenses able to reliably withstand such attacks against neural networks. However, adversarial training imposes a significant training overhead and scales poorly with model complexity and input dimension. In this paper, we propose Robust Representation Matching (RRM), a low-cost method to transfer the robustness of an adversarially trained model to a new model being trained for the same task irrespective of architectural differences. Inspired by student-teacher learning, our method introduces a novel training loss that encourages the student to learn the teacher's robust representations. Compared to prior works, RRM is superior with respect to both model performance and adversarial training time. On CIFAR-10, RRM trains a robust model ~1.8X faster than the state-of-the-art. Furthermore, RRM remains effective on higher-dimensional datasets. On Restricted-ImageNet, RRM trains a ResNet50 model ~18X faster than standard adversarial training.

Open Access Media

USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.

BibTeX
@inproceedings {279986,
author = {Pratik Vaishnavi and Kevin Eykholt and Amir Rahmati},
title = {Transferring Adversarial Robustness Through Robust Representation Matching},
booktitle = {31st USENIX Security Symposium (USENIX Security 22)},
year = {2022},
isbn = {978-1-939133-31-1},
address = {Boston, MA},
pages = {2083--2098},
url = {https://www.usenix.org/conference/usenixsecurity22/presentation/vaishnavi},
publisher = {USENIX Association},
month = aug
}

Presentation Video