Words or Vision：视觉-语言模型对文本有盲目信任吗？
Words or Vision: Do Vision-Language Models Have Blind Faith in Text?
AI总结：
本研究发现视觉-语言模型在模态不一致时存在过度信任文本的“盲目信任”现象，通过分析影响因素并提出监督微调与文本增强方法有效缓解该偏见，强调需平衡训练以提升多模态鲁棒性。
摘要：
视觉-语言模型（VLMs）在视觉中心任务中擅长整合视觉和文本信息，但它们对模态间不一致性的处理尚未得到充分探索。我们研究了在视觉中心环境中，面对视觉数据和多样化文本输入时，VLMs的模态偏好。通过对四个视觉中心任务引入文本变化，并评估十个视觉-语言模型（VLMs），我们发现了一种“对文本的盲目信任”现象：当出现不一致时，VLMs过度信任文本数据而非视觉数据，导致在文本损坏时性能显著下降，并引发安全担忧。我们分析了影响这种文本偏见的因素，包括指令提示、语言模型规模、文本相关性、token顺序以及视觉与文本确定性之间的相互作用。虽然某些因素（如扩大语言模型规模）能略微缓解文本偏见，但其他因素（如token顺序）可能因语言模型继承的位置偏见而加剧该问题。为解决此问题，我们探索了结合文本增强的监督微调，并证明其在减少文本偏见方面的有效性。此外，我们提供了理论分析，表明“对文本的盲目信任”现象可能源于训练期间纯文本数据与多模态数据的不平衡。我们的发现强调了在VLMs中需要平衡训练并谨慎考虑模态交互，以增强其在处理多模态数据不一致性时的鲁棒性和可靠性。
英文摘要：
Vision-Language Models (VLMs) excel in integrating visual and textual information for vision-centric tasks, but their handling of inconsistencies between modalities is underexplored. We investigate VLMs' modality preferences when faced with visual data and varied textual inputs in vision-centered settings. By introducing textual variations to four vision-centric tasks and evaluating ten Vision-Language Models (VLMs), we discover a \emph{``blind faith in text''} phenomenon: VLMs disproportionately trust textual data over visual data when inconsistencies arise, leading to significant performance drops under corrupted text and raising safety concerns. We analyze factors influencing this text bias, including instruction prompts, language model size, text relevance, token order, and the interplay between visual and textual certainty. While certain factors, such as scaling up the language model size, slightly mitigate text bias, others like token order can exacerbate it due to positional biases inherited from language models. To address this issue, we explore supervised fine-tuning with text augmentation and demonstrate its effectiveness in reducing text bias. Additionally, we provide a theoretical analysis suggesting that the blind faith in text phenomenon may stem from an imbalance of pure text and multi-modal data during training. Our findings highlight the need for balanced training and careful consideration of modality interactions in VLMs to enhance their robustness and reliability in handling multi-modal data inconsistencies.
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。
