arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2410.23728cs.CL
收藏

GigaCheck:通过以对象为中心的跨度定位检测LLM生成内容

GigaCheck: Detecting LLM-generated Content via Object-Centric Span Localization

SALUTEDEV LLC

Irina Tolstykh, Aleksandra Tsybina, Sergey Yakubson, Aleksandr Gordeev, Vladimir Dokholyan, Maksim Kuprashevich

AI总结:

GigaCheck提出双策略框架检测AI生成文本,利用细调LLM的表示学习进行作者身份识别,结合DETR-like模型与语言编码器实现精准定位,实验验证了方法的鲁棒性。

摘要:

随着LLM助手质量和传播的增加,生成内容迅速增长。在许多任务中,这些文本已与人类写作难以区分,生成质量持续提升。同时,检测方法进步速度慢于生成模型,难以防止生成AI技术的滥用。我们提出GigaCheck,一种双策略框架用于AI生成文本检测。在文档层面,利用细调LLM的表示学习实现高数据效率的作者身份识别。在跨度层面,引入新颖的结构适应,将生成文本段视为“对象”。通过整合DETR-like视觉模型与语言编码器,实现AI区间精准定位,有效将视觉对象检测的鲁棒性转移到文本领域。在三个分类和三个定位基准测试中,实验结果验证了方法的鲁棒性。共享细调的主干模型在两种场景中均表现出强准确性,突显了学习嵌入的泛化能力。此外,我们成功证明视觉检测架构如DETR不仅限于像素空间,能有效泛化到生成文本跨度的定位。为确保可重复性和促进进一步研究,我们公开发布源代码。

英文摘要:

With the increasing quality and spread of LLM assistants, the amount of generated content is growing rapidly. In many cases and tasks, such texts are already indistinguishable from those written by humans, and the quality of generation continues to increase. At the same time, detection methods are advancing more slowly than generation models, making it challenging to prevent misuse of generative AI technologies. We propose GigaCheck, a dual-strategy framework for AI-generated text detection. At the document level, we leverage the representation learning of fine-tuned LLMs to discern authorship with high data efficiency. At the span level, we introduce a novel structural adaptation that treats generated text segments as "objects." By integrating a DETR-like vision model with linguistic encoders, we achieve precise localization of AI intervals, effectively transferring the robustness of visual object detection to the textual domain. Experimental results across three classification and three localization benchmarks confirm the robustness of our approach. The shared fine-tuned backbone delivers strong accuracy in both scenarios, highlighting the generalization power of the learned embeddings. Moreover, we successfully demonstrate that visual detection architectures like DETR are not limited to pixel space, effectively generalizing to the localization of generated text spans. To ensure reproducibility and foster further research, we publicly release our source code.

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。

↑