paper-with-me

홈 › Papers

How does Watermarking Affect Visual Language Models in Document Understanding?

2025-04-01 · Chunxue Xu, Yiwei Wang, Bryan Hooi, Yujun Cai, Songze Li

Visual Language Models (VLMs) have become foundational models for document understanding tasks, widely used in the processing of complex multimodal documents across domains such as finance, law, and academia. However, documents often contain noise-like information, such as watermarks, which inevitably leads us to inquire: \emph{Do watermarks degrade the performance of VLMs in document understanding?} To address this, we propose a novel evaluation framework to investigate the effect of visible watermarks on VLMs performance. We takes into account various factors, including different types of document data, the positions of watermarks within documents and variations in watermark content. Our experimental results reveal that VLMs performance can be significantly compromised by watermarks, with performance drop rates reaching up to 36\%. We discover that \emph{scattered} watermarks cause stronger interference than centralized ones, and that \emph{semantic contents} in watermarks creates greater disruption than simple visual occlusion. Through attention mechanism analysis and embedding similarity examination, we find that the performance drops are mainly attributed to that watermarks 1) force widespread attention redistribution, and 2) alter semantic representation in the embedding space. Our research not only highlights significant challenges in deploying VLMs for document understanding, but also provides insights towards developing robust inference mechanisms on watermarked documents.

📄 PDF Abstract BibTeX arXiv:2504.01048

Code (0)

등록된 구현이 없습니다.

Tasks

document understanding

Methods 이 논문이 사용한 방법론

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Attention 설명 없음

Similar Papers 제목 키워드 기반

How Good is Post-Hoc Watermarking With Language Model Rephrasing?

2025-12-18 · Pierre Fernandez, Tom Sander, Hady Elsahar, Hongyan Chang 외 arxiv

Generation-time text watermarking embeds statistical signals into text for traceability of AI-generated content. We explore *post-hoc watermarking* where an LLM rewrites existing text while applying generation-time water…

SynthID-Image: Image watermarking at internet scale

2025-10-10 · Sven Gowal, Rudy Bunel, Florian Stimberg, David Stutz 외 arxiv

We introduce SynthID-Image, a deep learning-based system for invisibly watermarking AI-generated imagery. This paper documents the technical desiderata, threat models, and practical challenges of deploying such a system …

Testing the Effect of Code Documentation on Large Language Model Code Understanding

2024-04-03 · William Macke, Michael Doyle

Large Language Models (LLMs) have demonstrated impressive abilities in recent years with regards to code generation and understanding. However, little work has investigated how documentation and other code properties aff…

Code GenerationLanguage ModelingLanguage ModellingLarge Language Model

Cross-Attention Watermarking of Large Language Models

2024-01-12 · Folco Bertini Baldassini, Huy H. Nguyen, Ching-Chung Chang, Isao Echizen

A new approach to linguistic watermarking of language models is presented in which information is imperceptibly inserted into the output text while preserving its readability and original meaning. A cross-attention mecha…

Signature vs. Substance: Evaluating the Balance of Adversarial Resistance and Linguistic Quality in Watermarking Large Language Models

2025-08-11 · William Guo, Adaku Uchendu, Ana Smith arxiv

To mitigate the potential harms of Large Language Models (LLMs)generated text, researchers have proposed watermarking, a process of embedding detectable signals within text. With watermarking, we can always accurately de…