VIB-Probe：通过变分信息瓶颈检测和缓解视觉-语言模型中的幻觉
VIB-Probe: Detecting and Mitigating Hallucinations in Vision-Language Models via Variational Information Bottleneck
AI总结：
本文提出VIB-Probe框架，通过变分信息瓶颈理论检测和缓解视觉-语言模型中的幻觉，利用注意力头提取判别模式并过滤语义噪声，实验表明其在多个基准上表现优异。
摘要：
视觉-语言模型（VLMs）在多模态任务中展现了显著的进步，但仍然容易产生幻觉，即生成的文本与底层视觉内容偏离。现有的幻觉检测方法主要依赖输出logits或外部验证工具，往往忽视其内部机制。在本文中，我们研究了内部注意力头的输出，推测特定的头部承载了真实生成的主要信号。然而，直接探测这些高维状态具有挑战性，因为视觉-语言语法和噪声相互交织。为了解决这一问题，我们提出了VIB-Probe，一种利用变分信息瓶颈（VIB）理论的幻觉检测和缓解框架。我们的方法通过信息瓶颈原理在各层和头部中提取判别模式，并通过过滤掉语义噪声来实现。此外，通过利用我们VIB探针的梯度，我们识别出对幻觉具有强因果影响的注意力头部，并引入一种推理时的干预策略用于幻觉缓解。在多个基准上的广泛实验表明，VIB-Probe在两种设置中都显著优于现有基线。我们的代码将公开发布。
英文摘要：
Vision-Language Models (VLMs) have demonstrated remarkable progress in multimodal tasks, but remain susceptible to hallucinations, where generated text deviates from the underlying visual content. Existing hallucination detection methods primarily rely on output logits or external verification tools, often overlooking their internal mechanisms. In this work, we investigate the outputs of internal attention heads, postulating that specific heads carry the primary signals for truthful generation.However, directly probing these high-dimensional states is challenging due to the entanglement of visual-linguistic syntax and noise. To address this, we propose VIB-Probe, a novel hallucination detection and mitigation framework leveraging the Variational Information Bottleneck (VIB) theory. Our method extracts discriminative patterns across layers and heads while filtering out semantic nuisances through the information bottleneck principle. Furthermore, by leveraging the gradients of our VIB probe, we identify attention heads with strong causal influence on hallucinations and introduce an inference-time intervention strategy for hallucination mitigation. Extensive experiments across diverse benchmarks demonstrate that VIB-Probe significantly outperforms existing baselines in both settings. Our code will be made publicly available.
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。
