arXiv cs.CL· Hanxiu Zhang, Qiang Xu, Yue Zheng·· 2 天前AI 评分31
SELF：基于奇异值与特征值的 LLM 鲁棒指纹识别方法
SELF: A Robust Singular Value and Eigenvalue Approach for LLM Fingerprinting
AI 导读
研究者提出 SELF，一种基于 LLM 注意力权重内在特征的指纹识别方案，通过奇异值与特征值分解提取独特、可扩展且对变换不变的指纹，无需依赖输入即可抵御虚假声明攻击。实验显示 SELF 在保持高侵权检测准确率的同时，显著优于现有基线降低误报率，并对量化、剪枝和微调等下游修改表现出强鲁棒性。代码已公开。
正文 · 原文
Abstract:Safeguarding the Intellectual Property (IP) of Large Language Models (LLMs) against model theft represents a critical challenge in contemporary AI research. Although fingerprinting is essential for establishing forensic evidence for model derivation, existing methods are susceptible to false claim attacks, weight manipulations, or high false positive rates, undermining their utility. To overcome these limitations, we propose SELF, \rev{a novel intrinsic weight-based fingerprinting scheme tailored for robust IP forensics via adversarial model lineage verification}. SELF eliminates dependency on input and inherently resists false claims, achieving robust \rev{provenance detection} through unique, scalable and transformation-invariant fingerprint extraction via singular value and eigenvalue decomposition of LLM attention weights. Experimental results demonstrate that SELF maintains high IP infringement detection accuracy while significantly outperforming state-of-the-art baselines in reducing false positives and showing strong robustness against various downstream modifications, including quantization, pruning, and fine-tuning attacks. Our code is available at this https URL.
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2512.03620 [cs.CR] |
| (or arXiv:2512.03620v2 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2512.03620 arXiv-issued DOI via DataCite |
Submission history
From: Hanxiu Zhang [view email]
[v1]
Wed, 3 Dec 2025 09:53:47 UTC (2,244 KB)
[v2]
Tue, 27 Oct 2026 03:24:28 UTC (1,393 KB)
来源：arXiv cs.CL · arxiv.org