arXiv cs.CL· Jinmo Han, Jimin Hong, Chanyeong Moon, Ju Yeon Kang, Seonuk Kim, Nam Soo Kim·· 12 小时前AI 评分26
MD-ProTector：用多数据驱动原型定位检测 LLM 生成文本
MD-ProTector: Positioning Multiple Data-Driven Prototypes for LLM-Generated Text Detection
AI 导读
MD-ProTector 通过编码器嵌入空间中的多个可训练原型为同一类内不同文本分组提供独立决策边界，并引入 Prototype Positioning loss 区分类级结构与类内差异。在覆盖领域、生成模型、语言和对抗变化的三个大规模基准五项设置上，该方法在 MAGE CDCM 和 RAID 上取得最高 AvgRec，并在 RAID 上取得最高 AUROC 与最低 FPR95。
正文 · 原文
Abstract:As LLM-generated content becomes more sophisticated, detection systems for distinguishing those texts from human-written text must operate at scale while handling diverse writing styles, domains, languages, and generator models. Input-only encoder detectors are suitable for practical deployment setting, but standard binary classification supplies only the class label and does not explicitly organize the substantial variation within either class. We propose MD-ProTector, which represents each class with multiple trainable reference vectors in the encoder embedding space, referred to as prototypes. These prototypes provide separate decision boundaries for different groups of texts within the same class. However, adding multiple prototypes alone does not determine which variation each prototype should represent. MD-ProTector addresses this problem with Prototype Positioning loss, which separates class-level structure from the within-class variation that differentiates individual prototypes. Evaluated across five settings from three large-scale benchmarks covering domain, generator, language, and adversarial variation, MD-ProTector achieves the highest AvgRec on MAGE CDCM and RAID and the highest AUROC and lowest FPR95 on RAID among the compared encoder-based methods.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2608.10459 [cs.CL] |
| (or arXiv:2608.10459v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2608.10459 arXiv-issued DOI via DataCite |
Submission history
From: Jinmo Han [view email]
[v1]
Tue, 11 Aug 2026 04:14:10 UTC (1,763 KB)
[v2]
Tue, 3 Nov 2026 20:02:49 UTC (1,766 KB)
来源：arXiv cs.CL · arxiv.org