UMD: Unsupervised Model Detection for X2X Backdoor Attacks
This addresses a critical security vulnerability in AI systems by enabling unsupervised detection of complex backdoor attacks, which is a significant advancement over existing supervised methods.
The paper tackles the problem of detecting X2X backdoor attacks in deep neural networks, where multiple source classes are misclassified to arbitrary target classes, and proposes UMD, an unsupervised method that achieves detection accuracy improvements of 17%, 4%, and 8% over state-of-the-art supervised detectors on CIFAR-10, GTSRB, and Imagenette datasets.
Backdoor (Trojan) attack is a common threat to deep neural networks, where samples from one or more source classes embedded with a backdoor trigger will be misclassified to adversarial target classes. Existing methods for detecting whether a classifier is backdoor attacked are mostly designed for attacks with a single adversarial target (e.g., all-to-one attack). To the best of our knowledge, without supervision, no existing methods can effectively address the more general X2X attack with an arbitrary number of source classes, each paired with an arbitrary target class. In this paper, we propose UMD, the first Unsupervised Model Detection method that effectively detects X2X backdoor attacks via a joint inference of the adversarial (source, target) class pairs. In particular, we first define a novel transferability statistic to measure and select a subset of putative backdoor class pairs based on a proposed clustering approach. Then, these selected class pairs are jointly assessed based on an aggregation of their reverse-engineered trigger size for detection inference, using a robust and unsupervised anomaly detector we proposed. We conduct comprehensive evaluations on CIFAR-10, GTSRB, and Imagenette dataset, and show that our unsupervised UMD outperforms SOTA detectors (even with supervision) by 17%, 4%, and 8%, respectively, in terms of the detection accuracy against diverse X2X attacks. We also show the strong detection performance of UMD against several strong adaptive attacks.