大模型对齐与安全 - arXivDaily 专题

2508.04105 2025-08-07 cs.AI 57%

Towards Transparent AI Grading: Semantic Entropy as a Signal for Human-AI Disagreement

Karrtik Iyer, Manikandan Ravikiran, Prasanna Pendse, Shayan Mohanty

机构 * Thoughtworks AI Research Labs（Thoughtworks人工智能研究实验室）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.AI

2508.04005 2025-08-07 cs.LG 57%

Decoupled Contrastive Learning for Federated Learning

Hyungbin Kim, Incheol Baek, Yon Dohn Chung

专题命中幻觉与事实性：alignment(abstract)；分类 cs.LG

2502.12591 2025-08-06 cs.CV cs.CL 57%

CutPaste&Find: Efficient Multimodal Hallucination Detector with Visual-aid Knowledge Base

Cong-Duy Nguyen, Xiaobao Wu, Duc Anh Vu, Shuai Zhao, Thong Nguyen, Anh Tuan Luu

机构 * Nanyang Technological University, Singapore（南洋理工大学）； National University of Singapore, Singapore（国立新加坡大学）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

2508.02419 2025-08-05 cs.CV cs.CL 57%

Modality Bias in LVLMs: Analyzing and Mitigating Object Hallucination via Attention Lens

Haohan Zheng, Zhenguo Zhang

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

2508.01678 2025-08-05 cs.CV cs.AI 57%

Cure or Poison? Embedding Instructions Visually Alters Hallucination in Vision-Language Models

Zhaochen Wang, Yiwei Wang, Yujun Cai

机构 * The University of Queensland（昆士兰大学）； University of California, Merced（加州大学默塞德分校）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.AI

Comments Work in progress

2507.23736 2025-08-01 stat.ML cs.LG 57%

DICOM De-Identification via Hybrid AI and Rule-Based Framework for Scalable, Uncertainty-Aware Redaction

Kyle Naddeo, Nikolas Koutsoubis, Rahul Krish, Ghulam Rasool, Nidhal Bouaynaya, Tony OSullivan, Raj Krish

机构 * Rowan University（罗文大学）； Moffitt Cancer Center（莫菲特癌症中心）； University of South Florida（佛罗里达州立大学）； Impact Business Information Solutions, Inc（Impact商务信息解决方案公司）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.LG

Comments 15 pages, 6 figures,

2507.21406 2025-07-30 cs.AI 57%

Shapley Uncertainty in Natural Language Generation

Meilin Zhu, Gaojie Jin, Xiaowei Huang, Lijun Zhang

机构 * University of Exeter（埃克塞特大学）； University of Liverpool（利物浦大学）； ISCAS（国际信息科学协会）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.AI

2507.16771 2025-07-23 cs.LG stat.AP stat.ML 57%

A Partitioned Sparse Variational Gaussian Process for Fast, Distributed Spatial Modeling

Michael Grosskopf, Kellin Rumsey, Ayan Biswas, Earl Lawrence

专题命中幻觉与事实性：alignment(abstract)；分类 cs.LG

2505.19073 2025-07-22 cs.CL 57%

Towards Harmonized Uncertainty Estimation for Large Language Models

Rui Li, Jing Long, Muge Qi, Heming Xia, Lei Sha, Peiyi Wang, Zhifang Sui

机构 * Peking University（北京大学）； The Hong Kong Polytechnic University（香港理工大学）； Beihang University（北航）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.CL

Comments ACL 2025

2507.14782 2025-07-22 stat.ML cs.LG math-ph math.MP stat.CO 57%

Uncertainty Quantification for Machine Learning-Based Prediction: A Polynomial Chaos Expansion Approach for Joint Model and Input Uncertainty Propagation

Xiaoping Du

机构 * School of Mechanical Engineering（机械工程学院）； Purdue University（普渡大学）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.LG

Comments This manuscript has been submitted to Multidisciplinary and Structural Optimization

2407.14540 2025-07-18 cs.SE cs.AI cs.CR 57%

Risks of ignoring uncertainty propagation in AI-augmented security pipelines

Emanuele Mezzi, Aurora Papotti, Fabio Massacci, Katja Tuma

机构 * Vrije Universiteit Amsterdam（阿姆斯特丹自由大学）； University of Trento（特伦托大学）； Eindhoven University of Technology（埃因霍温理工大学）

专题命中幻觉与事实性：safety(abstract)；分类 cs.AI

Comments Accepted for publication in Risk Analysis: An International Journal

2506.23146 2025-07-15 cs.CL 57%

Learning-to-Context Slope: Evaluating In-Context Learning Effectiveness Beyond Performance Illusions

Dingzriui Wang, Xuanliang Zhang, Keyan Xu, Qingfu Zhu, Wanxiang Che, Yang Deng

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

2507.06571 2025-07-10 cs.CL 57%

Enhancing Food-Domain Question Answering with a Multimodal Knowledge Graph: Hybrid QA Generation and Diversity Analysis

Srihari K B, Pushpak Bhattacharyya

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

2505.12182 2025-07-09 cs.CL 57%

Truth Neurons

Haohang Li, Yupeng Cao, Yangyang Yu, Jordan W. Suchow, Zining Zhu

机构 * Stevens Institute of Technology（史蒂文斯理工学院）

专题命中幻觉与事实性：safety(abstract)；分类 cs.CL

2505.05163 2025-07-08 cs.CV cs.LG 57%

Probabilistic Embeddings for Frozen Vision-Language Models: Uncertainty Quantification with Gaussian Process Latent Variable Models

Aishwarya Venkataramanan, Paul Bodesheim, Joachim Denzler

机构 * Computer Vision Group, Friedrich Schiller University Jena（计算机视觉组，费迪里奇·施勒尔大学耶纳）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.LG

Comments UAI 2025, 22 pages

2412.10435 2025-07-04 cs.CV cs.AI 57%

COEF-VQ: Cost-Efficient Video Quality Understanding through a Cascaded Multimodal LLM Framework

Xin Dong, Sen Jia, Ming Rui Wang, Yan Li, Zhenheng Yang, Bingfeng Deng, Hongyu Xiong

机构 * ByteDance Inc.（字节跳动公司）

专题命中幻觉与事实性：safety(abstract)；分类 cs.AI

2409.17815 2025-07-03 cs.AI 57%

DREAMS: A python framework for Training Deep Learning Models on EEG Data with Model Card Reporting for Medical Applications

Rabindra Khadka, Pedro G Lind, Anis Yazidi, Asma Belhadi

机构 * Department of Computer Science, OsloMet -- Oslo Metropolitan University（计算机科学系，奥斯陆Met大学）； Simula Research Laboratory, Numerical Analysis and Scientific Computing（Simula研究实验室，数值分析与科学计算）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.AI

2506.23164 2025-07-01 cs.RO cs.AI 57%

Mode Collapse Happens: Evaluating Critical Interactions in Joint Trajectory Prediction Models

Maarten Hugenholtz, Anna Meszaros, Jens Kober, Zlatan Ajanovic

机构 * Cognitive Robotics Department, Delft University of Technology（代尔夫特理工大学认知机器人学系）； Computer Science Department, RWTH Aachen University（亚琛工业大学计算机科学系）

专题命中幻觉与事实性：safety(abstract)；分类 cs.AI

Comments 12 pages, 8 figures, submitted to a journal

2506.19513 2025-06-25 cs.CV cs.LG 57%

Visual hallucination detection in large vision-language models via evidential conflict

Tao Huang, Zhekun Liu, Rui Wang, Yang Zhang, Liping Jing

机构 * Beijing Key Lab of Traffic Data Mining（北京交通数据挖掘与具身智能重点实验室）； State Key Laboratory of Advanced Rail Autonomous Operation（先进轨道交通自主运行国家重点实验室）； School of Computer Science and Technology（计算机科学与技术学院）； Beijing Jiaotong University（北京交通大学）； School of Automation and Intelligence（自动化与智能学院）； School of Electronic and Information Engineering（电子与信息工程学院）

专题命中幻觉与事实性：safety(abstract)；分类 cs.LG

Journal ref International Journal of Approximate Reasoning, Volume 186, November 2025, Article 109507

2506.18559 2025-06-24 cs.AI cs.LO 57%

T-CPDL: A Temporal Causal Probabilistic Description Logic for Developing Logic-RAG Agent

Hong Qing Yu

机构 * University of Derby（德比大学）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.AI

2406.00958 2025-06-24 cs.LG cs.CV 57%

Navigating Conflicting Views: Harnessing Trust for Learning

Jueqing Lu, Wray Buntine, Yuanyuan Qi, Joanna Dipnall, Belinda Gabbe, Lan Du

机构 * Department of Data Science \& AI, Monash University ； School of Public Health ； Preventive Medicine, Monash University ； College of Engineering ； Computer Science, VinUniversity

专题命中幻觉与事实性：alignment(abstract)；分类 cs.LG

Journal ref Proceedings of the 42nd International Conference on Machine Learning (ICML 2025)

2506.15889 2025-06-23 cs.CL 57%

Entropy-Driven Pre-Tokenization for Byte-Pair Encoding

Yifan Hu, Frank Liang, Dachuan Zhao, Jonathan Geuter, Varshini Reddy, Craig W. Schmidt, Chris Tanner

机构 * Harvard University, Cambridge, MA（哈佛大学）； Kempner Institute, Allston, MA（凯姆纳研究所）； Kensho Technologies, Cambridge, MA（Kensho技术公司）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

2506.09272 2025-06-12 cs.LG stat.ML 57%

G-Sim: Generative Simulations with Large Language Models and Gradient-Free Calibration

Samuel Holt, Max Ruiz Luyten, Antonin Berthon, Mihaela van der Schaar

机构 * University of Cambridge（剑桥大学）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.LG

Comments Accepted at the 42nd International Conference on Machine Learning (ICML 2025). 9 pages, 3 figures

Journal ref Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025

2506.08422 2025-06-12 cs.AI 57%

Transforming Expert Knowledge into Scalable Ontology via Large Language Models

Ikkei Itoku, David Theil, Evelyn Eichelsdoerfer Uehara, Sreyoshi Bhaduri, Junnosuke Kuroda, Toshi Yumoto, Alex Gil, Natalie Perez, Rajesh Cherukuri, Naumaan Nayyar

机构 * Amazon New York, USA（亚马逊纽约分公司）； Amazon Arlington, USA（亚马逊阿灵顿分公司）； Amazon Seattle, USA（亚马逊西雅图分公司）； Amazon Honolulu, USA（亚马逊檀香山分公司）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.AI

2412.13949 2025-06-11 cs.CL cs.CV 57%

Cracking the Code of Hallucination in LVLMs with Vision-aware Head Divergence

Jinghan He, Kuan Zhu, Haiyun Guo, Junfeng Fang, Zhenglin Hua, Yuheng Jia, Ming Tang, Tat-Seng Chua, Jinqiao Wang

机构 * Foundation Model Research Center, Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所基础模型研究中心）； School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）； National University of Singapore（新加坡国立大学）； Southeast University（东南大学）； Wuhan AI Research（武汉人工智能研究所）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

Comments ACL2025

2506.07227 2025-06-10 cs.CV cs.CL 57%

Hallucination at a Glance: Controlled Visual Edits and Fine-Grained Multimodal Learning

Tianyi Bai, Yuxuan Fan, Jiantao Qiu, Fupeng Sun, Jiayi Song, Junlin Han, Zichen Liu, Conghui He, Wentao Zhang, Binhang Yuan

机构 * HKUST（香港科技大学）； Shanghai AI Lab（上海人工智能实验室）； Peking University（北京大学）； HKUST(GZ)（香港科技大学（广州））； Oxford University（牛津大学）； Imperial College London（伦敦帝国理工学院）

专题命中幻觉与事实性：alignment(abstract)；分类 cs.CL

2506.05826 2025-06-09 cs.LG 57%

Learning Along the Arrow of Time: Hyperbolic Geometry for Backward-Compatible Representation Learning

Ngoc Bui, Menglin Yang, Runjin Chen, Leonardo Neves, Mingxuan Ju, Rex Ying, Neil Shah, Tong Zhao

专题命中幻觉与事实性：alignment(abstract)；分类 cs.LG

2506.01616 2025-06-03 cs.AI 57%

MLA-Trust: Benchmarking Trustworthiness of Multimodal LLM Agents in GUI Environments

Xiao Yang, Jiawei Chen, Jun Luo, Zhengwei Fang, Yinpeng Dong, Hang Su, Jun Zhu

机构 * Department of Computer Science & Technology, Institute for AI, BNRist Center, THBI Lab, Tsinghua-Bosch Joint Center for ML（计算机科学与技术系、人工智能研究院、BNRist中心、THBI实验室、清华-博世联合机器学习中心）； Shanghai Key Lab. of Multidimensional Info. Processing（上海多维信息处理重点实验室）

专题命中幻觉与事实性：safety(abstract)；分类 cs.AI

2505.24649 2025-06-02 cs.CV cs.AI 57%

BIMA: Bijective Maximum Likelihood Learning Approach to Hallucination Prediction and Mitigation in Large Vision-Language Models

Huu-Thien Tran, Thanh-Dat Truong, Khoa Luu

机构 * CVIU Lab, University of Arkansas（CVIU实验室，阿肯色大学）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.AI

Comments CVPRW 2025, 8 pages, 4 figures

2505.24377 2025-06-02 cs.CL 57%

LLM Inference Enhanced by External Knowledge: A Survey

Yu-Hsuan Lin, Qian-Hui Chen, Yi-Jie Cheng, Jia-Ren Zhang, Yi-Hung Liu, Liang-Yu Hsia, Yun-Nung Chen

机构 * National Taiwan University（国立台湾大学）

专题命中幻觉与事实性：trustworthy(abstract)；分类 cs.CL

1. 幻觉与事实性 1756 篇

Towards Transparent AI Grading: Semantic Entropy as a Signal for Human-AI Disagreement

Decoupled Contrastive Learning for Federated Learning

CutPaste&Find: Efficient Multimodal Hallucination Detector with Visual-aid Knowledge Base

Modality Bias in LVLMs: Analyzing and Mitigating Object Hallucination via Attention Lens

Cure or Poison? Embedding Instructions Visually Alters Hallucination in Vision-Language Models

DICOM De-Identification via Hybrid AI and Rule-Based Framework for Scalable, Uncertainty-Aware Redaction

Shapley Uncertainty in Natural Language Generation

A Partitioned Sparse Variational Gaussian Process for Fast, Distributed Spatial Modeling

Towards Harmonized Uncertainty Estimation for Large Language Models

Uncertainty Quantification for Machine Learning-Based Prediction: A Polynomial Chaos Expansion Approach for Joint Model and Input Uncertainty Propagation

Risks of ignoring uncertainty propagation in AI-augmented security pipelines

Learning-to-Context Slope: Evaluating In-Context Learning Effectiveness Beyond Performance Illusions

Enhancing Food-Domain Question Answering with a Multimodal Knowledge Graph: Hybrid QA Generation and Diversity Analysis

Truth Neurons

Probabilistic Embeddings for Frozen Vision-Language Models: Uncertainty Quantification with Gaussian Process Latent Variable Models

COEF-VQ: Cost-Efficient Video Quality Understanding through a Cascaded Multimodal LLM Framework

DREAMS: A python framework for Training Deep Learning Models on EEG Data with Model Card Reporting for Medical Applications

Mode Collapse Happens: Evaluating Critical Interactions in Joint Trajectory Prediction Models

Visual hallucination detection in large vision-language models via evidential conflict

T-CPDL: A Temporal Causal Probabilistic Description Logic for Developing Logic-RAG Agent

Navigating Conflicting Views: Harnessing Trust for Learning

Entropy-Driven Pre-Tokenization for Byte-Pair Encoding

G-Sim: Generative Simulations with Large Language Models and Gradient-Free Calibration

Transforming Expert Knowledge into Scalable Ontology via Large Language Models

Cracking the Code of Hallucination in LVLMs with Vision-aware Head Divergence

Hallucination at a Glance: Controlled Visual Edits and Fine-Grained Multimodal Learning

Learning Along the Arrow of Time: Hyperbolic Geometry for Backward-Compatible Representation Learning

MLA-Trust: Benchmarking Trustworthiness of Multimodal LLM Agents in GUI Environments

BIMA: Bijective Maximum Likelihood Learning Approach to Hallucination Prediction and Mitigation in Large Vision-Language Models

LLM Inference Enhanced by External Knowledge: A Survey