权衡导航：视觉语言模型中零样本对抗鲁棒性防御策略的综合研究
Navigating the Trade-off: A Synthesis of Defensive Strategies for Zero-Shot Adversarial Robustness in Vision-Language Models
AI总结：
本报告综合八篇论文，分析视觉语言模型零样本对抗鲁棒性中鲁棒性与泛化的权衡，对比对抗微调与免训练防御，并展望混合防御与对抗预训练方向。
摘要：
本报告综合了八篇关于视觉语言模型（如CLIP）零样本对抗鲁棒性的重要论文。该领域的一个核心挑战是增强对抗鲁棒性与保持模型零样本泛化能力之间的固有权衡。我们分析了两种主要的防御范式：对抗微调（AFT），它修改模型参数；以及免训练/测试时防御，它保留模型参数。我们追溯了从对齐保持方法（TeCoA）到嵌入空间重构（LAAT, TIMA），以及从输入启发式方法（AOM, TTC）到潜在空间净化（CLIPure）的演变过程。最后，我们指出了关键挑战和未来方向，包括混合防御策略和对抗预训练。
英文摘要：
This report synthesizes eight seminal papers on the zero-shot adversarial robustness of vision-language models (VLMs) like CLIP. A central challenge in this domain is the inherent trade-off between enhancing adversarial robustness and preserving the model's zero-shot generalization capabilities. We analyze two primary defense paradigms: Adversarial Fine-Tuning (AFT), which modifies model parameters, and Training-Free/Test-Time Defenses, which preserve them. We trace the evolution from alignment-preserving methods (TeCoA) to embedding space re-engineering (LAAT, TIMA), and from input heuristics (AOM, TTC) to latent-space purification (CLIPure). Finally, we identify key challenges and future directions including hybrid defense strategies and adversarial pre-training.
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。
