DetectGPT 是否充分利用了扰动？将选择性扰动桥接到微调对比学习检测器会更好
Does DetectGPT Fully Utilize Perturbation? Bridging Selective Perturbation to Fine-tuned Contrastive Learning Detector would be Better
AI总结：
针对 DetectGPT 随机扰动引入噪声且依赖阈值的问题，本文提出微调检测器 Pecola，通过选择性扰动保留重要 token 并进行加权多对对比学习，在四个公开数据集上平均准确率超越 SOTA 1.20%。
摘要：
大语言模型（LLMs）日益增长的生成能力引发了人们对滥用的担忧，这要求自动化的机器生成文本检测器。DetectGPT 是一种基于度量的零样本检测器，首次引入了扰动并展现出显著的性能提升。然而，在 DetectGPT 中，随机扰动策略可能引入噪声，并且 logit 回归依赖于阈值，损害了其对单个或小批量输入的泛化性和适用性。因此，我们提出了一种新颖的微调检测器 Pecola，通过在选择性扰动上进行对比学习，将基于度量的方法和微调方法桥接起来。选择性策略在扰动过程中保留重要 token，并为多对对比学习赋予权重。实验表明，Pecola 在四个公开数据集上的平均准确率比最先进（SOTA）方法高出 1.20%。此外，我们进一步分析了该方法的有效性、鲁棒性和泛化性。
英文摘要：
The burgeoning generative capabilities of large language models (LLMs) have raised growing concerns about abuse, demanding automatic machine-generated text detectors. DetectGPT, a zero-shot metric-based detector, first introduces perturbation and shows great performance improvement. However, in DetectGPT, the random perturbation strategy could introduce noise, and logit regression depends on the threshold, harming the generalizability and applicability of individual or small-batch inputs. Hence, we propose a novel fine-tuned detector, Pecola, bridging metric-based and fine-tuned methods by contrastive learning on selective perturbation. Selective strategy retains important tokens during perturbation and weights for multi-pair contrastive learning. The experiments show that Pecola outperforms the state-of-the-art (SOTA) by 1.20% in accuracy on average on four public datasets. And we further analyze the effectiveness, robustness, and generalization of the method.
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。
