Extreme Miscalibration and the Illusion of Adversarial Robustness

Vyas Raina, Samson Tan, Volkan Cevher, Aditya Rawal, Sheng Zha, George Karypis


Abstract
Deep learning-based Natural Language Processing (NLP) models are vulnerable to adversarial attacks, where small perturbations can cause a model to misclassify. Adversarial Training (AT) is often used to increase model robustness. However, we have discovered an intriguing phenomenon: deliberately or accidentally miscalibrating models masks gradients in a way that interferes with adversarial attack search methods, giving rise to an apparent increase in robustness. We show that this observed gain in robustness is an illusion of robustness (IOR), and demonstrate how an adversary can perform various forms of test-time temperature calibration to nullify the aforementioned interference and allow the adversarial attack to find adversarial examples. Hence, we urge the NLP community to incorporate test-time temperature scaling into their robustness evaluations to ensure that any observed gains are genuine. Finally, we show how the temperature can be scaled during training to improve genuine robustness.
Anthology ID:
2024.acl-long.137
Volume:
Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
Month:
August
Year:
2024
Address:
Bangkok, Thailand
Editors:
Lun-Wei Ku, Andre Martins, Vivek Srikumar
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2500–2525
Language:
URL:
https://aclanthology.org/2024.acl-long.137
DOI:
10.18653/v1/2024.acl-long.137
Bibkey:
Cite (ACL):
Vyas Raina, Samson Tan, Volkan Cevher, Aditya Rawal, Sheng Zha, and George Karypis. 2024. Extreme Miscalibration and the Illusion of Adversarial Robustness. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2500–2525, Bangkok, Thailand. Association for Computational Linguistics.
Cite (Informal):
Extreme Miscalibration and the Illusion of Adversarial Robustness (Raina et al., ACL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.acl-long.137.pdf