arXiv cs.CL· Yu-Shin Huang, Chao Tian, Krishna Narayanan·· 11 小时前AI 评分20
大语言模型多比特生成水印：最坏情况误报约束下的最优方案
Optimal Multi-bit Generative Watermarking Schemes Under Worst-Case False-Alarm Constraints
AI 导读
针对大语言模型多比特生成水印，研究证明此前声称达到有限 token 区间漏检概率下界的方案实际次优，并提出两种新的编解码构造真正达到该下界，从而完整刻画最优多比特水印性能。新方法将水印设计建模为线性规划，并给出实现最优性的结构条件，同时分析了旧方案的失效机制并比较两种新方案的权衡。
正文 · 原文
Abstract:This paper considers the problem of multi-bit generative watermarking for large language models under a worst-case false-alarm constraint. Prior work established a lower bound on the achievable miss-detection probability in the finite-token regime and proposed a scheme claimed to achieve this bound. We show, however, that the proposed scheme is in fact suboptimal. We then develop two new encoding-decoding constructions that attain the previously established lower bound, thereby completely characterizing the optimal multi-bit watermarking performance. Our approach formulates the watermark design problem as a linear program and derives the structural conditions under which optimality can be achieved. In addition, we identify the failure mechanism of the previous construction and compare the tradeoffs between the two proposed schemes.
| Comments: | 42 pages, 8 tables |
| Subjects: | Information Theory (cs.IT); Computation and Language (cs.CL) |
| Cite as: | arXiv:2604.08759 [cs.IT] |
| (or arXiv:2604.08759v2 [cs.IT] for this version) | |
| https://doi.org/10.48550/arXiv.2604.08759 arXiv-issued DOI via DataCite |
Submission history
From: Yu-Shin Huang [view email]
[v1]
Thu, 9 Apr 2026 20:46:24 UTC (37 KB)
[v2]
Fri, 30 Oct 2026 16:19:40 UTC (40 KB)
来源：arXiv cs.CL · arxiv.org