arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.19947cs.CRcs.AI
收藏

静默触发器,明显足迹:后门隐身性的三空间测量研究

Quiet Triggers, Loud Footprints: A Tri-Space Measurement Study of Backdoor Stealthiness

Radboud University(拉德堡德大学);University of Zagreb Faculty of Electrical Engineering and Computing(萨格勒布大学电气工程和计算学院)

Behrad Tajalli, Henk Berendsen, Xiaoyun Xu, Stjepan Picek

AI总结:

本文针对图像分类任务开展后门隐身性的三空间测量研究,引入两个诊断指标,发现现有攻击难以在输入、特征、参数三空间均实现低足迹,为后续评估提供实用指南。

摘要:

针对深度神经网络的后门攻击通常通过攻击成功率和良性准确率进行评估,但其隐身性的定义和测量却不一致。现有研究往往仅考察隐身性的一个可观测方面,如触发器的不可感知性、特征空间的可分性、参数空间的集中度,或规避特定防御措施,这导致了碎片化的认知:一种攻击可能在某一指标下表现隐身,却在其他方面留下明显足迹。本文针对图像分类任务开展后门隐身性的三空间测量研究,区分了潜在隐身性(即区分后门系统与良性系统的理想但不可观测的难度)与输入、特征、参数三个空间中可测量的足迹。在该框架下,我们使用一套广泛的足迹指标,评估了四种数据集和三种模型架构上的代表性后门攻击。我们的结果表明,隐身性并非单一整体:在某一空间中表现静默的攻击,往往在另一空间中表现明显,且现有攻击很少能在三个空间中均实现持续较小的足迹。我们进一步分析了隐身性指标间的一致性与不一致性,研究了足迹测量与攻击成功率的相互作用。基于现有测量中观察到的盲区,我们引入了两个诊断指标:类特定戴维斯-布ldin指数(CDBI),用于捕捉类条件特征重叠;以及TAC-UCLC乘积(TUP),用于识别既对触发器有响应又对下游有影响的通道。这些指标旨在补充而非替代更广泛的测量套件。最后,我们为未来的隐身性评估提供了实用指南。

英文摘要:

Backdoor attacks on deep neural networks are commonly evaluated by attack success rate and benign accuracy, yet their stealthiness remains inconsistently defined and measured. Existing work often studies only one observable aspect of stealthiness, such as trigger imperceptibility, feature-space separability, parameter-space concentration, or evasion of particular defenses. This creates a fragmented picture: an attack may appear stealthy under one metric while leaving conspicuous footprints elsewhere. This paper presents a tri-space measurement study of backdoor stealthiness in image classification. We distinguish latent stealthiness---the ideal but unobservable difficulty of distinguishing a backdoored system from a benign one---from measurable footprints in three spaces: input, feature, and parameter space. Under this framework, we evaluate representative backdoor attacks across four datasets and three model architectures using a broad suite of footprint metrics. Our results show that stealthiness is not monolithic: attacks that are quiet in one space are often loud in another, and existing attacks rarely achieve consistently small footprints across all three spaces. We further analyze agreement and disagreement among stealthiness metrics and study how footprint measurements interact with attack success rate. Based on observed blind spots in existing measurements, we introduce two diagnostic metrics: Class-Specific Davies--Bouldin Index (CDBI), which captures class-conditional feature overlap, and TAC--UCLC Product (TUP), which identifies channels that are both trigger-responsive and downstream influential. These metrics are intended to augment, not replace, the broader measurement suite. Finally, we provide practical guidance for future stealthiness evaluations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑