无需对抗样本，通过对比网络检测模型中的恶意输入。
Pulling Back the Curtain: Unsupervised Adversarial Detection via Contrastive Auxiliary Networks
- 在目标模型中间层嵌入对比辅助网络，学习区分正常与对抗特征。
- 在多个数据集和模型上，对四种攻击方法的F1分数均优于现有无监督方法。
- 适合需要提升深度学习系统安全性的研究人员和工程师使用。
深度学习模型广泛应用于安全关键场景，但仍易受对抗攻击——即难以察觉的扰动，可显著降低模型性能。传统防御机制主要聚焦于增强模型鲁棒性或独立检测对抗输入。本文提出一种无监督对抗检测方法U-CAN（Contrastive Auxiliary Networks），通过在目标模型选定的中间层嵌入辅助网络，利用投影层和基于ArcFace的线性层，从辅助特征表示中揭示对抗行为，无需依赖对抗样本。在CIFAR-10、Mammals及ImageNet子集等多个数据集，以及ResNet-50、VGG-16和ViT等模型架构上进行的全面实验表明，该方法在四种不同攻击方式下均优于现有无监督检测技术，取得更优的F1分数。所提框架为提升深度学习系统的安全性和可靠性提供了可扩展且高效的方法。
原文摘要 · Abstract (English)
Deep learning models are widely employed in safety-critical applications yet remain susceptible to adversarial attacks -- imperceptible perturbations that can significantly degrade model performance. Conventional defense mechanisms predominantly focus on either enhancing model robustness or detecting adversarial inputs independently. In this work, we propose an Unsupervised adversarial detection via Contrastive Auxiliary Networks (U-CAN) to uncover adversarial behavior within auxiliary feature representations, without the need for adversarial examples. U-CAN is embedded within selected intermediate layers of the target model. These auxiliary networks, comprising projection layers and ArcFace-based linear layers, refine feature representations to more effectively distinguish between benign and adversarial inputs. Comprehensive experiments across multiple datasets (CIFAR-10, Mammals, and a subset of ImageNet) and architectures (ResNet-50, VGG-16, and ViT) demonstrate that our method surpasses existing unsupervised adversarial detection techniques, achieving superior F1 scores against four distinct attack methods. The proposed framework provides a scalable and effective solution for enhancing the security and reliability of deep learning systems.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。