MAA：针对视觉-语言预训练模型的细致对抗攻击
MAA: Meticulous Adversarial Attack against Vision-Language Pre-trained Models
AI总结：
本文提出细致对抗攻击（MAA），通过缩放与滑动裁剪（RScrop）和多粒度相似性破坏（MGSD）策略，利用模型无关特征优化对抗图像，显著提升了针对视觉-语言预训练模型攻击的有效性与可迁移性。
摘要：
当前用于评估多模态任务中视觉-语言预训练（VLP）模型鲁棒性的对抗攻击方法存在可迁移性有限的问题，即为特定模型设计的攻击往往难以在不同模型间有效泛化，限制了其在更广泛范围内评估鲁棒性的效用。这主要归因于对模型特定特征和区域的过度依赖，尤其是在图像模态中。本文提出了一种优雅且高效的方法，称为细致对抗攻击（MAA），以充分利用单个样本的模型无关特征和漏洞，从而实现增强的泛化能力并降低模型依赖。MAA通过开发一种新颖的缩放与滑动裁剪（RScrop）技术，并结合多粒度相似性破坏（MGSD）策略，强调对对抗图像的细粒度优化。在多种VLP模型、多个基准数据集以及多种下游任务上的大量实验表明，MAA显著提升了对抗攻击的有效性和可迁移性。研究还进行了大规模性能分析，为各种模型配置的有效性提供了深入见解，指导该领域的未来发展。
英文摘要：
Current adversarial attacks for evaluating the robustness of vision-language pre-trained (VLP) models in multi-modal tasks suffer from limited transferability, where attacks crafted for a specific model often struggle to generalize effectively across different models, limiting their utility in assessing robustness more broadly. This is mainly attributed to the over-reliance on model-specific features and regions, particularly in the image modality. In this paper, we propose an elegant yet highly effective method termed Meticulous Adversarial Attack (MAA) to fully exploit model-independent characteristics and vulnerabilities of individual samples, achieving enhanced generalizability and reduced model dependence. MAA emphasizes fine-grained optimization of adversarial images by developing a novel resizing and sliding crop (RScrop) technique, incorporating a multi-granularity similarity disruption (MGSD) strategy. Extensive experiments across diverse VLP models, multiple benchmark datasets, and a variety of downstream tasks demonstrate that MAA significantly enhances the effectiveness and transferability of adversarial attacks. A large cohort of performance studies is conducted to generate insights into the effectiveness of various model configurations, guiding future advancements in this domain.
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。
