arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2409.01151cs.CVcs.LG
收藏

通过无参数表示对齐理解多模态幻觉

Understanding Multimodal Hallucination with Parameter-Free Representation Alignment

Wangxuan Institute of Computer Technology, Peking University(北京大学王选计算机研究所);National Key Laboratory of General Artificial Intelligence(通用人工智能全国重点实验室)

Yueqian Wang, Jianxin Liang, Yuxuan Wang, Huishuai Zhang, Dongyan Zhao

AI总结:

本文提出无参数表示对齐度量Pfram,通过测量图像表示与人类标注的对齐程度,揭示其与多模态大语言模型物体幻觉的强相关性,并分析模块作用、指令影响及编码器改进。

摘要:

幻觉是多模态大语言模型(MLLMs)中的常见问题,但其底层原理仍未被充分理解。本文研究了 MLLMs 中哪些组件会导致物体幻觉。为了在完全避免除图像表示本身之外的所有其他因素影响的情况下分析图像表示,我们提出了一种无参数表示对齐度量(Pfram),该度量可以在不需要额外训练参数的情况下测量任意两个表示系统之间的相似性。值得注意的是,Pfram 还可以评估神经表示系统与人类表示系统(由图像的真实标注表示)之间的对齐程度。通过评估与物体标注的对齐程度,我们证明该度量在广泛的先进 MLLMs 中与物体幻觉表现出强且一致的相关性,涵盖各种模型架构和规模。此外,利用该度量,我们探索了与 MLLMs 中图像表示相关的其他关键问题,例如不同模块的作用、文本指令的影响,以及包括使用替代视觉编码器在内的潜在改进。我们的代码可在以下地址获取:https://github.com/yellow-binary-tree/Pfram。

英文摘要:

Hallucination is a common issue in Multimodal Large Language Models (MLLMs), yet the underlying principles remain poorly understood. In this paper, we investigate which components of MLLMs contribute to object hallucinations. To analyze image representations while completely avoiding the influence of all other factors other than the image representation itself, we propose a parametric-free representation alignment metric (Pfram) that can measure the similarities between any two representation systems without requiring additional training parameters. Notably, Pfram can also assess the alignment of a neural representation system with the human representation system, represented by ground-truth annotations of images. By evaluating the alignment with object annotations, we demonstrate that this metric shows strong and consistent correlations with object hallucination across a wide range of state-of-the-art MLLMs, spanning various model architectures and sizes. Furthermore, using this metric, we explore other key issues related to image representations in MLLMs, such as the role of different modules, the impact of textual instructions, and potential improvements including the use of alternative visual encoders. Our code is available at: https://github.com/yellow-binary-tree/Pfram.

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。

↑