InterPol 论文提出通过插值偏好学习去匿名化 LM Arena 模型
InterPol: De-anonymizing LM Arena via Interpolated Preference Learning
AI 导读
论文提出 INTERPOL 框架，利用模型插值合成难负样本并结合自适应课程学习，识别 LM Arena 中匿名回复的目标模型，克服 TF-IDF 等统计特征难以区分风格相近模型的局限。实验显示其识别准确率显著优于现有基线，并通过 Arena 对战数据上的排名操纵模拟量化了现实威胁。
正文 · 原文
Abstract:Strict anonymity of model responses is a key for the reliability of voting-based leaderboards, such as LM Arena. While prior studies have attempted to compromise this assumption using simple statistical features like TF-IDF or bag-ofwords, these methods often lack the discriminative power to distinguish between stylistically similar or within-family models. To overcome these limitations and expose the severity of vulnerability, we introduce INTERPOL, a model-driven identification framework that learns to distinguish target models from others using interpolated preference data. Specifically, INTERPOL captures deep stylistic patterns that superficial statistical features miss by synthesizing hard negative samples through model interpolation and employing an adaptive curriculum learning strategy. Extensive experiments demonstrate that INTERPOL significantly outperforms existing baselines in identification accuracy. Furthermore, we quantify the real-world threat of our findings through ranking manipulation simulations on Arena battle data.
| Comments: | EMNLP 2026 (SAC highlight award) |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2603.15220 [cs.AI] |
| (or arXiv:2603.15220v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2603.15220 arXiv-issued DOI via DataCite |
Submission history
From: Minsung Cho [view email]
[v1]
Mon, 16 Mar 2026 12:56:39 UTC (1,785 KB)
[v2]
Wed, 23 Sep 2026 11:14:53 UTC (813 KB)
[v3]
Thu, 29 Oct 2026 05:54:40 UTC (812 KB)
来源：arXiv cs.AI · arxiv.org