双曲安全感知视觉-语言模型
Hyperbolic Safety-Aware Vision-Language Models
AI总结：
本文提出 HySAC，利用双曲空间的蕴含层级建模安全与不安全图文对的非对称关系，使 CLIP 兼具不安全内容分类、灵活检索及安全重定向能力。
摘要：
解决 CLIP 等视觉-语言模型检索不安全内容的问题，是其走向真实世界集成的重要一步。现有工作依赖遗忘技术，试图擦除模型对不安全概念的知识。尽管这类方法能有效减少不良输出，但遗忘会限制模型区分安全与不安全内容的能力。本文提出一种新方法，利用双曲空间固有的层次结构属性，从“遗忘”范式转向“感知”范式。我们建议将安全与不安全内容编码为一个蕴含层级，使二者位于双曲空间的不同区域。我们的 HySAC（Hyperbolic Safety-Aware CLIP，双曲安全感知 CLIP）采用蕴含损失函数，对安全与不安全图文对之间的层级和非对称关系进行建模。由于标准视觉-语言模型依赖欧氏嵌入，这种建模在其中无法有效发挥作用；而 HySAC 由此获得对不安全内容的感知能力，使其既可作为多模态不安全分类器，也可作为灵活的内容检索器，并支持将不安全查询动态重定向到更安全的替代内容，或保留原始输出。大量实验表明，该方法不仅增强了安全识别能力，还为视觉-语言模型中的内容审核建立了更具适应性和可解释性的框架。源代码可在 https://github.com/aimagelab/HySAC 获取。
英文摘要：
Addressing the retrieval of unsafe content from vision-language models such as CLIP is an important step towards real-world integration. Current efforts have relied on unlearning techniques that try to erase the model's knowledge of unsafe concepts. While effective in reducing unwanted outputs, unlearning limits the model's capacity to discern between safe and unsafe content. In this work, we introduce a novel approach that shifts from unlearning to an awareness paradigm by leveraging the inherent hierarchical properties of the hyperbolic space. We propose to encode safe and unsafe content as an entailment hierarchy, where both are placed in different regions of hyperbolic space. Our HySAC, Hyperbolic Safety-Aware CLIP, employs entailment loss functions to model the hierarchical and asymmetrical relations between safe and unsafe image-text pairs. This modelling, ineffective in standard vision-language models due to their reliance on Euclidean embeddings, endows the model with awareness of unsafe content, enabling it to serve as both a multimodal unsafe classifier and a flexible content retriever, with the option to dynamically redirect unsafe queries toward safer alternatives or retain the original output. Extensive experiments show that our approach not only enhances safety recognition but also establishes a more adaptable and interpretable framework for content moderation in vision-language models. Our source code is available at https://github.com/aimagelab/HySAC.
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。
