Fortify the Guardian, Not the Treasure: Resilient Adversarial Detectors
This addresses the challenge of securing machine learning systems against sophisticated attacks for practitioners, though it is incremental as it builds on existing adversarial training methods.
The paper tackles the problem of making adversarial detectors robust against adaptive attacks while preserving classifier accuracy, achieving significant improvements in detection accuracy on CIFAR-10 and SVHN datasets without compromising clean performance.
This paper presents RADAR-Robust Adversarial Detection via Adversarial Retraining-an approach designed to enhance the robustness of adversarial detectors against adaptive attacks, while maintaining classifier performance. An adaptive attack is one where the attacker is aware of the defenses and adapts their strategy accordingly. Our proposed method leverages adversarial training to reinforce the ability to detect attacks, without compromising clean accuracy. During the training phase, we integrate into the dataset adversarial examples, which were optimized to fool both the classifier and the adversarial detector, enabling the adversarial detector to learn and adapt to potential attack scenarios. Experimental evaluations on the CIFAR-10 and SVHN datasets demonstrate that our proposed algorithm significantly improves a detector's ability to accurately identify adaptive adversarial attacks -- without sacrificing clean accuracy.