CeTAD：迈向视觉语言模型中的可认证毒性感知距离
CeTAD: Towards Certified Toxicity-Aware Distance in Vision Language Models
AI总结：
针对视觉语言模型易受视觉越狱攻击的鲁棒性问题，提出CeTAD通用可认证防御框架，含新型语义距离度量、回归认证方法及特征空间防御，可提供正式鲁棒性保证，提升模型韧性与可信度。
摘要：
大型视觉语言模型（VLM）的最新进展已在各类视觉理解任务中展现出显著成效，但这类模型针对越狱攻击的鲁棒性仍是一项待解决的挑战。本研究提出一个通用可认证防御框架，为VLM提供针对潜在视觉越狱攻击的严格防护。首先，我们提出一种新型距离度量指标，用于量化恶意响应与预期响应之间的语义差异，能够捕捉传统基于余弦相似度的度量方法常忽略的细微差别。随后，我们设计了一种回归认证方法，采用随机平滑技术，即使在黑盒设置下也能为对抗性扰动和结构性扰动提供正式的鲁棒性保证。作为补充，我们的特征空间防御方法将噪声分布（如高斯分布、拉普拉斯分布）引入潜在嵌入中，以同时抵御像素级和结构级扰动。研究结果凸显了基于正式理论的整合策略在构建更具韧性、更可信的VLM方面的潜力。
英文摘要：
Recent advances in large vision-language models (VLMs) have demonstrated remarkable success across a wide range of visual understanding tasks. However, the robustness of these models against jailbreak attacks remains an open challenge. In this work, we propose a universal certified defence framework to safeguard VLMs rigorously against potential visual jailbreak attacks. First, we proposed a novel distance metric to quantify semantic discrepancies between malicious and intended responses, capturing subtle differences often overlooked by conventional cosine similarity-based measures. Then, we devise a regressed certification approach that employs randomized smoothing to provide formal robustness guarantees against both adversarial and structural perturbations, even under black-box settings. Complementing this, our feature-space defence introduces noise distributions (e.g., Gaussian, Laplacian) into the latent embeddings to safeguard against both pixel-level and structure-level perturbations. Our results highlight the potential of a formally grounded, integrated strategy toward building more resilient and trustworthy VLMs.
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。
