突破生成器障碍:面向通用化AI-文本检测的解耦表示
Breaking the Generator Barrier: Disentangled Representation for Generalizable AI-Text Detection
AI总结:
本文提出解耦框架,通过紧凑潜在编码和扰动正则化提升AI-文本检测的泛化能力,在MAGE基准上取得显著提升。
摘要:
随着大型语言模型生成的文本越来越接近人类写作,区分AI生成内容与人类写作的细微线索变得越来越难以捕捉。依赖生成器特定的特征本质上是不稳定的,因为新模型迅速出现,降低了此类捷径的鲁棒性。将未见过的生成器泛化是AI-文本检测中的核心挑战。为此,我们提出一个逐步结构化的框架,将AI检测语义与生成器感知特征解耦。这通过紧凑的潜在编码促进语义最小性,随后基于扰动的正则化减少残余纠缠,最后通过判别适应阶段对齐表示与任务目标。在MAGE基准上,涵盖20个代表性LLM的7个类别,实验显示在状态-of-the-art方法上取得一致改进,达到最高24.2%的准确率提升和26.2%的F1改进。值得注意的是,随着训练生成器多样性的增加,性能持续提升,证实了在开放集场景中的强大可扩展性和泛化能力。我们的源代码将在https://github.com/PuXiao06/DRGD上公开。
英文摘要:
As large language models (LLMs) generate text that increasingly resembles human writing, the subtle cues that distinguish AI-generated content from human-written content become increasingly challenging to capture. Reliance on generator-specific artifacts is inherently unstable, since new models emerge rapidly and reduce the robustness of such shortcuts. This generalizes unseen generators as a central and challenging problem for AI-text detection. To tackle this challenge, we propose a progressively structured framework that disentangles AI-detection semantics from generator-aware artifacts. This is achieved through a compact latent encoding that encourages semantic minimality, followed by perturbation-based regularization to reduce residual entanglement, and finally a discriminative adaptation stage that aligns representations with task objectives. Experiments on MAGE benchmark, covering 20 representative LLMs across 7 categories, demonstrate consistent improvements over state-of-the-art methods, achieving up to 24.2% accuracy gain and 26.2% F1 improvement. Notably, performance continues to improve as the diversity of training generators increases, confirming strong scalability and generalization in open-set scenarios. Our source code will be publicly available at https://github.com/PuXiao06/DRGD.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。