arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2501.09425cs.CVcs.CL
收藏

视觉-语言模型并不理解否定

Vision-Language Models Do Not Understand Negation

Institution1(机构1);Institution2(机构2)

Kumail Alhamoud, Shaden Alshammari, Yonglong Tian, Guohao Li, Philip Torr, Yoon Kim, Marzyeh Ghassemi

AI总结:

本文提出NegBench评估VLM的否定理解能力,发现现有模型常接近随机水平,并通过合成否定字幕微调CLIP显著提升检索与选择题表现。

摘要:

许多实际的视觉-语言应用需要模型理解否定,例如在使用自然语言检索包含某些对象但不包含其他对象的图像时。尽管视觉-语言模型(VLM)已通过大规模训练取得进展,但它们理解否定的能力仍未得到充分研究。本研究探讨的问题是:当前VLM对否定的理解程度如何?我们引入了NegBench,这是一个新基准,旨在跨18种任务变体和涵盖图像、视频及医学数据集的7.9万个样本评估否定理解。该基准包含两项核心任务,用于在多样的多模态场景中评估否定理解:带否定的检索,以及带否定字幕的多项选择题。我们的评估表明,现代VLM在否定处理上明显吃力,往往只达到随机水平。为解决这些不足,我们探索了一种以数据为中心的方法,在包含数百万条否定字幕的大规模合成数据集上微调CLIP模型。结果表明,该方法可使否定查询的召回率提升10%,并使带否定字幕的多项选择题准确率提高28%。

英文摘要:

Many practical vision-language applications require models that understand negation, e.g., when using natural language to retrieve images which contain certain objects but not others. Despite advancements in vision-language models (VLMs) through large-scale training, their ability to comprehend negation remains underexplored. This study addresses the question: how well do current VLMs understand negation? We introduce NegBench, a new benchmark designed to evaluate negation understanding across 18 task variations and $79$k examples spanning image, video, and medical datasets. The benchmark consists of two core tasks designed to evaluate negation understanding in diverse multimodal settings: Retrieval with Negation and Multiple Choice Questions with Negated Captions. Our evaluation reveals that modern VLMs struggle significantly with negation, often performing at chance level. To address these shortcomings, we explore a data-centric approach wherein we finetune CLIP models on large-scale synthetic datasets containing millions of negated captions. We show that this approach can result in a 10% increase in recall on negated queries and a 28% boost in accuracy on multiple-choice questions with negated captions.

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。

↑