解构幻觉：正交语义投影实现鲁棒可解释性
Disentangling Hallucinations: Orthogonal Semantic Projection for Robust Interpretability
AI总结：
针对视觉语言模型解释中的语义幻觉问题，提出线性语义归因（LSA）理论框架，并引入正交语义投影（OSP）方法，通过正交化查询向量消除共享特征干扰，最小化幻觉。
摘要：
随着视觉语言模型在安全关键型应用中的部署日益增多，其解释的可信度变得至关重要。视觉语言模型的可解释人工智能（XAI）方法常常遭受语义幻觉，即当输入错误的文本描述时（例如，提示“猫”却高亮显示狗），归因图仍会突出显示显著的图像区域。尽管这个问题普遍存在，但文献中缺乏对XAI方法和CLIP嵌入的正式数学分析。我们证明，这种现象并非特定于单一架构，而是高维嵌入空间中线性语义泄漏的基本后果。我们提出了一个统一的理论框架——线性语义归因（LSA），该框架泛化于多种判别方法。我们引入了OSP，一种利用OMP残差性质的几何干预方法，用于将独特的语义信号与共享概念分离。我们从理论上证明并实验表明，OSP通过将查询向量与干扰概念正交化，最小化幻觉，使归因模型对共享特征“失明”，同时保持对正确提示的保真度。我们的代码可在 https://github.com/emirhanbilgic/Orthogonal-Semantic-Projection 获取。
英文摘要：
As Vision-Language Models are increasingly deployed in safety-critical applications, the trustworthiness of their explanations becomes crucial. Explainable AI (XAI) methods for Vision-Language Models often suffer from semantic hallucination, where attribution maps highlight prominent image regions even when prompted with incorrect text descriptions (e.g., highlighting a dog when prompted ``cat''). Although this problem is widespread, a formal mathematical analysis of XAI methods and CLIP embeddings is largely missing in the literature. We demonstrate that this phenomenon is not specific to a single architecture but is a fundamental consequence of Linear Semantic Leakage in high-dimensional embedding spaces. We propose a unified theoretical framework, Linear Semantic Attribution (LSA), which generalizes across discriminative methods. We introduce OSP, a geometric intervention that utilizes the residual property of OMP to disentangle unique semantic signals from shared concepts. We prove theoretically and demonstrate empirically that OSP minimizes hallucination by orthogonalizing the query vector against distractor concepts, rendering the attribution model blind to shared features while preserving fidelity for correct prompts. Our code is available at: https://github.com/emirhanbilgic/Orthogonal-Semantic-Projection
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。
