arXiv 2601.192452026-02-19cs.AIcs.LG

超越领域检测：SpikeScore用于跨领域幻觉检测

Beyond In-Domain Detection: SpikeScore for Cross-Domain Hallucination Detection

University of Technology Sydney（技术科技大学）；University of Wisconsin-Madison（威斯康星大学麦迪逊分校）

Yongxin Deng, Zhen Fang, Sharon Li, Ling Chen

发表信息：: In Proceedings of the Fourteenth International Conference on Learning Representations (ICLR 2026)

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出SpikeScore方法，通过量化多轮对话中的不确定性波动，实现跨领域幻觉检测的高效识别与高泛化性能。

摘要：

幻觉检测对于在现实世界中部署大语言模型（LLMs）至关重要。现有的幻觉检测方法在训练和测试数据来自同一领域时表现强劲，但在跨领域泛化方面却表现不佳。在本文中，我们研究了一个重要但被忽视的问题，称为可泛化幻觉检测（GHD），其目标是利用单一领域的数据训练幻觉检测器，同时确保在不同相关领域中具有鲁棒的性能。在研究GHD时，我们模拟了遵循LLM初始响应的多轮对话，并观察到一个有趣的现象：幻觉引发的多轮对话在不同领域中普遍表现出比事实对话更大的不确定性波动。基于这一现象，我们提出了一种新的评分SpikeScore，它量化了多轮对话中的突发波动。通过理论分析和实证验证，我们证明SpikeScore在幻觉和非幻觉响应之间实现了强大的跨领域分离性。在多个LLM和基准测试中的实验表明，基于SpikeScore的检测方法在跨领域泛化方面优于代表性的基线方法，并超过了先进的泛化导向方法，验证了我们的方法在跨领域幻觉检测中的有效性。

英文摘要：

Hallucination detection is critical for deploying large language models (LLMs) in real-world applications. Existing hallucination detection methods achieve strong performance when the training and test data come from the same domain, but they suffer from poor cross-domain generalization. In this paper, we study an important yet overlooked problem, termed generalizable hallucination detection (GHD), which aims to train hallucination detectors on data from a single domain while ensuring robust performance across diverse related domains. In studying GHD, we simulate multi-turn dialogues following LLMs' initial response and observe an interesting phenomenon: hallucination-initiated multi-turn dialogues universally exhibit larger uncertainty fluctuations than factual ones across different domains. Based on the phenomenon, we propose a new score SpikeScore, which quantifies abrupt fluctuations in multi-turn dialogues. Through both theoretical analysis and empirical validation, we demonstrate that SpikeScore achieves strong cross-domain separability between hallucinated and non-hallucinated responses. Experiments across multiple LLMs and benchmarks demonstrate that the SpikeScore-based detection method outperforms representative baselines in cross-domain generalization and surpasses advanced generalization-oriented methods, verifying the effectiveness of our method in cross-domain hallucination detection.

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。