通过多重检验原则检测大语言模型中的幻觉
Principled Detection of Hallucinations in Large Language Models via Multiple Testing
AI总结：
本文提出通过多重检验方法系统整合多个评估分数，实现受控的幻觉检测，验证了方法在不同模型和数据集上的鲁棒性。
摘要：
虽然大语言模型（LLMs）已成为解决各种任务的强大基础模型，但它们也被发现容易产生幻觉，即生成听起来自信但实际上是错误或无意义的响应。现有的幻觉检测器提出了多种经验评分规则，但其性能在不同模型和数据集上变化较大，难以确定在实践中应依赖哪些或作为可靠的检测器。在本文中，我们将检测幻觉的问题公式化为假设检验问题，并与机器学习模型中的分布外检测问题进行类比。我们随后提出了一种受多重检验启发的方法，通过符合性p值系统整合多个评估分数，实现校准的检测并控制误报率。在广泛模型和数据集上的大量实验验证了我们方法对最新方法的鲁棒性。
英文摘要：
While Large Language Models (LLMs) have emerged as powerful foundational models to solve a variety of tasks, they have also been shown to be prone to hallucinations, i.e., generating responses that sound confident but are actually incorrect or even nonsensical. Existing hallucination detectors propose a wide range of empirical scoring rules, but their performance varies across models and datasets, and it is hard to determine which ones to rely on in practice or to treat as a reliable detector. In this work, we formulate the problem of detecting hallucinations as a hypothesis testing problem and draw parallels with the problem of out-of-distribution detection in machine learning models. We then propose a multiple-testing-inspired method that systematically aggregates multiple evaluation scores via conformal p-values, enabling calibrated detection with controlled false alarm rate. Extensive experiments across diverse models and datasets validate the robustness of our approach against state-of-the-art methods.
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。
