arXiv 2605.009392026-05-12cs.LGcs.AI

从平面事实到尖锐幻觉：通过梯度敏感性检测顽固错误

From Flat Facts to Sharp Hallucinations: Detecting Stubborn Errors via Gradient Sensitivity

School of Intelligent Manufacturing Ecosystem, Xi’an Jiaotong-Liverpool University, People’s Republic of China（智能制造生态系统学院，西安交通大学-利物浦大学，中华人民共和国）；Department of Computer Science, University of Liverpool, United Kingdom（计算机科学系，利物浦大学，英国）；Faculty of Engineering and Technology, Sunway University, Malaysia（工程与技术学院，Sunway大学，马来西亚）；School of Robotics, Xi’an Jiaotong-Liverpool University, People’s Republic of China（机器人学院，西安交通大学-利物浦大学，中华人民共和国）

Yee Zhing Liew, Andrew Huey Ping Tan, Anwar P. P. Abdul Majeed

备注：: Accepted to ICML 2026. Camera-ready version

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出通过梯度敏感性检测顽固错误，利用嵌入扰动梯度敏感性（EPGS）区分稳定知识与不稳定记忆，实验显示其优于熵基和表示基方法。

摘要：

传统幻觉检测在

英文摘要：

Traditional hallucination detection fails on "Stubborn Hallucinations" - errors where LLMs are confidently wrong. We propose a geometric solution: Embedding-Perturbed Gradient Sensitivity (EPGS). We hypothesize that while robust facts reside in flat minima, stubborn hallucinations sit in sharp minima, supported by brittle memorization. EPGS detects this sharpness by perturbing input embeddings with Gaussian noise and measuring the resulting spike in gradient magnitude. This acts as an efficient proxy for the Hessian spectrum, differentiating stable knowledge from unstable memorization. Our experiments show that EPGS significantly outperforms entropy-based and representation-based baselines, providing a robust signal for detecting high-confidence factual errors.

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。