新方法提高跨领域AI文本检测的鲁棒性

报道来源 [4]

1. arXiv cs.CL TIER_1 English(EN) · Mohamed Mady, Johannes Reschke, Bj\"orn Schuller · 2026-05-06 04:00

   用于跨领域和生成器的鲁棒AI文本检测的特征增强Transformer

   arXiv:2605.03969v1 Announce Type: new Abstract: AI-generated text is nowadays produced at scale across domains and heterogeneous generation pipelines, making robustness to distribution shift a central requirement for supervised binary detectors. We train transformer-based detecto…

2. arXiv cs.CL TIER_1 English(EN) · Björn Schuller · 2026-05-05 16:52

   面向跨领域和生成器的鲁棒AI文本检测的特征增强Transformer

   AI-generated text is nowadays produced at scale across domains and heterogeneous generation pipelines, making robustness to distribution shift a central requirement for supervised binary detectors. We train transformer-based detectors on HC3 PLUS and calibrate a single decision t…

3. Hugging Face Daily Papers TIER_1 English(EN) · 2026-05-05 16:52

   面向跨领域和生成器的鲁棒AI文本检测的特征增强Transformer

   AI-generated text is nowadays produced at scale across domains and heterogeneous generation pipelines, making robustness to distribution shift a central requirement for supervised binary detectors. We train transformer-based detectors on HC3 PLUS and calibrate a single decision t…

4. arXiv cs.CL TIER_1 English(EN) · Priyadarshan Narayanasamy, Swastik Agrawal, Klint Faber, Fardina Fathmiul Alam · 2026-05-05 04:00

   超越困惑度：字符分布签名与用于AI文本检测的MDTA基准

   arXiv:2605.01647v1 Announce Type: new Abstract: Training-free AI text detection methods primarily rely on model log-probabilities, achieving strong performance through approaches like Binoculars and DNA-DetectLLM. However, these methods face a fundamental ceiling as models are op…