大模型推理能力 - arXivDaily 专题

2503.09095 2025-11-18 cs.CR cs.CV 50%

Backdooring CLIP through Concept Confusion

Lijie Hu, Junchi Liao, Weimin Lyu, Shaopeng Fu, Tianhao Huang, Shu Yang, Guimin Hu, Di Wang

机构 * Provable Responsible AI and Data Analytics (PRADA) Lab（可证负责任人工智能与数据分析实验室）； King Abdullah University of Science and Technology（卡塔尔国王科技大学）； University of Electronic Science and Technology of China（电子科技大学）； Stony Brook University（石溪大学）； University of Virginia（弗吉尼亚大学）； University of Copenhagen（哥本哈根大学）

专题命中复杂问题求解：reasoning(abstract)

2511.12982 2025-11-18 cs.CR cs.CV 50%

SafeGRPO: Self-Rewarded Multimodal Safety Alignment via Rule-Governed Policy Optimization

Xuankun Rong, Wenke Huang, Tingfeng Wang, Daiguo Zhou, Bo Du, Mang Ye

机构 * School of Computer Science, Wuhan University（武汉大学计算机学院）； MiLM Plus, Xiaomi Inc.（小米公司）

专题命中复杂问题求解：reasoning(abstract)

2509.24037 2025-11-17 cs.CR 50%

Automated Vulnerability Validation and Verification: A Large Language Model Approach

Alireza Lotfi, Charalampos Katsis, Elisa Bertino

专题命中复杂问题求解：reasoning(abstract)

2510.16823 2025-11-13 cs.SE cs.CR 50%

When AI Takes the Wheel: Security Analysis of Framework-Constrained Program Generation

Yue Liu, Zhenchang Xing, Shidong Pan, Chakkrit Tantithamthavorn

专题命中复杂问题求解：reasoning(abstract)

2303.16975 2025-11-13 cs.CV 50%

EgoTV: Egocentric Task Verification from Natural Language Task Descriptions

Rishi Hazra, Brian Chen, Akshara Rai, Nitin Kamra, Ruta Desai

机构 * Örebro University（奥雷布罗大学）； Meta

专题命中复杂问题求解：reasoning(abstract)

Comments Accepted at ICCV 2023

2511.08521 2025-11-12 cs.CV 50%

UniVA: Universal Video Agent towards Open-Source Next-Generation Video Generalist

Zhengyang Liang, Daoan Zhang, Huichi Zhou, Rui Huang, Bobo Li, Yuechen Zhang, Shengqiong Wu, Xiaohan Wang, Jiebo Luo, Lizi Liao, Hao Fei

机构 * Singapore Management University（新加坡管理大学）； University of Rochester（罗切斯特大学）； University College London（伦敦大学学院）； National University of Singapore（新加坡国立大学）； The Chinese University of Hong Kong（香港中文大学）； Stanford University（斯坦福大学）

专题命中复杂问题求解：reasoning(abstract)

Comments Technical Report. 24 figures, 37 pages. Website: https://univa.online/

2511.08334 2025-11-12 cs.CV 50%

Empowering DINO Representations for Underwater Instance Segmentation via Aligner and Prompter

Zhiyang Chen, Chen Zhang, Hao Fang, Runmin Cong

专题命中复杂问题求解：reasoning(abstract)

Comments AAAI 2026

2511.06888 2025-11-12 cs.CV 50%

A Two-Stage System for Layout-Controlled Image Generation using Large Language Models and Diffusion Models

Jan-Hendrik Koch, Jonas Krumme, Konrad Gadzicki

机构 * Cognitive Neuroinformatics University of Bremen（认知神经信息学不莱梅大学）

专题命中复杂问题求解：planning(abstract)

Comments 12 pages, 5 figures

2504.06958 2025-11-12 cs.CV 50%

VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning

Xinhao Li, Ziang Yan, Desen Meng, Lu Dong, Xiangyu Zeng, Yinan He, Yali Wang, Yu Qiao, Yi Wang, Limin Wang

机构 * Shanghai AI Laboratory（上海人工智能实验室）； Nanjing University（南京大学）； Zhejiang University（浙江大学）； University of Science and Technology of China（中国科学技术大学）； Shanghai Innovation Institute（上海创新研究院）； Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（中国科学院深圳先进技术研究所）

专题命中复杂问题求解：reasoning(abstract)

2511.06360 2025-11-11 cs.CV 50%

AesTest: Measuring Aesthetic Intelligence from Perception to Production

Guolong Wang, Heng Huang, Zhiqiang Zhang, Wentian Li, Feilong Ma, Xin Jin

机构 * University of International Business and Economics（国际商务经济大学）； University of Science and Technology of China（中国科学技术大学）； Huawei Technologies Co., Ltd（华为技术有限公司）； Beijing Electronic Science and Technology Institute（北京电子科技研究所）； Beijing Institute for General Artificial Intelligence（北京通用人工智能研究院）

专题命中复杂问题求解：reasoning(abstract)

Comments 10 pages, 9 figures

2511.05855 2025-11-11 cs.RO 50%

Gentle Manipulation Policy Learning via Demonstrations from VLM Planned Atomic Skills

Jiayu Zhou, Qiwei Wu, Jian Li, Zhe Chen, Xiaogang Xiong, Renjing Xu

专题命中复杂问题求解：planning(abstract)

Comments Accepted for the 40th Annual AAAI Conference on Artificial Intelligence (2026)

2502.20263 2025-11-11 cs.CV 50%

Vector-Quantized Vision Foundation Models for Object-Centric Learning

Rongzhen Zhao, Vivienne Wang, Juho Kannala, Joni Pajarinen

机构 * Aalto University（阿alto大学）； University of Oulu（奥卢大学）

专题命中复杂问题求解：reasoning(abstract)

Comments Accepted to ACM MM 2025

2511.03844 2025-11-07 cs.MA 50%

ASAP: an Agentic Solution to Auto-optimize Performance of Large-Scale LLM Training

Yuran Ding, Xinwei Chen, Xiaofan Zhang, Zongwei Zhou

专题命中复杂问题求解：reasoning(abstract)

Comments This work has been accepted to Workshop on ML for Systems at NeurIPS 2025

2511.03478 2025-11-06 cs.HC 50%

SVG Decomposition for Enhancing Large Multimodal Models Visualization Comprehension: A Study with Floor Plans

Jeongah Lee, Ali Sarvghad

专题命中复杂问题求解：reasoning(abstract)

Comments 10 pages, 2 figures

2511.02182 2025-11-05 cs.CV 50%

Pinpointing Trigger Moment for Grounded Video QA: Enhancing Spatio-temporal Grounding in Multimodal Large Language Models

Jinhwan Seo, Yoonki Cho, Junhyug Noh, Sung-eui Yoon

机构 * KAIST（韩国科学技术院）； Ewha Womans University（成均馆大学）

专题命中复杂问题求解：reasoning(abstract)

Comments 1st place winner of Grounded Videoqa track at the ICCV2025 Perception Test

2511.01617 2025-11-04 cs.CV cs.IR 50%

Vote-in-Context: Turning VLMs into Zero-Shot Rank Fusers

Mohamed Eltahir, Ali Habibullah, Lama Ayash, Tanveer Hussain, Naeemullah Khan

机构 * King Abdullah University of Science and Technology (KAUST)（卡布斯大学）； King Khalid University (KKU)（国王 Khalid 大学）； Edge Hill University（埃德希尔大学）

专题命中复杂问题求解：reasoning(abstract)

2511.01274 2025-11-04 cs.CV 50%

PRevivor: Reviving Ancient Chinese Paintings using Prior-Guided Color Transformers

Tan Tang, Yanhong Wu, Junming Gao, Yingcai Wu

专题命中复杂问题求解：reasoning(abstract)

2511.00805 2025-11-04 cs.IR 50%

REaR: Retrieve, Expand and Refine for Effective Multitable Retrieval

Rishita Agarwal, Himanshu Singhal, Peter Baile Chen, Manan Roy Choudhury, Dan Roth, Vivek Gupta

专题命中复杂问题求解：reasoning(abstract)

Comments 13 pages, 2 figures, 8 tables

2510.26027 2025-10-31 cs.CV 50%

Enhancing Temporal Understanding in Video-LLMs through Stacked Temporal Attention in Vision Encoders

Ali Rasekh, Erfan Bagheri Soula, Omid Daliran, Simon Gottschalk, Mohsen Fayyaz

机构 * Leibniz University Hannover（莱比锡大学汉诺威分校）； L3S Research Center（L3S研究中心）； Microsoft（微软公司）

专题命中复杂问题求解：reasoning(abstract)

Comments Accepted to NeurIPS 2025

2510.22396 2025-10-28 cs.CR 50%

PortGPT: Towards Automated Backporting Using Large Language Models

Zhaoyang Li, Zheng Yu, Jingyi Song, Meng Xu, Yuxuan Luo, Dongliang Mu

专题命中复杂问题求解：reasoning(abstract)

Comments Accepted by IEEE S&P 2026

2510.20622 2025-10-24 cs.CV 50%

SeViCES: Unifying Semantic-Visual Evidence Consensus for Long Video Understanding

Yuan Sheng, Yanbin Hao, Chenxu Li, Shuo Wang, Xiangnan He

机构 * University of Science and Technology of China（中国科学技术大学）； Hefei University of Technology（合肥工业大学）

专题命中复杂问题求解：reasoning(abstract)

2510.20223 2025-10-24 cs.CR cs.MM 50%

Beyond Text: Multimodal Jailbreaking of Vision-Language and Audio Models through Perceptually Simple Transformations

Divyanshu Kumar, Shreyas Jena, Nitin Aravind Birur, Tanay Baswa, Sahil Agarwal, Prashanth Harshangi

专题命中复杂问题求解：reasoning(abstract)

2510.17150 2025-10-23 cs.RO 50%

OmniVIC: A Self-Improving Variable Impedance Controller with Vision-Language In-Context Learning for Safe Robotic Manipulation

Heng Zhang, Wei-Hsing Huang, Gokhan Solak, Arash Ajoudani

机构 * Human-Robot Interfaces and Interaction Lab, Istituto Italiano di Tecnologia, Genoa, Italy（人机交互实验室，意大利理工学院，热那亚）； Georgia Institute of Technology, Atlanta, USA（佐治亚理工学院，美国亚特兰大）

专题命中复杂问题求解：reasoning(abstract)

Comments Code, video and RAG dataset are available at \url{https://sites.google.com/view/omni-vic}

2510.16880 2025-10-23 cs.CE 50%

Chem-R: Learning to Reason as a Chemist

Weida Wang, Benteng Chen, Di Zhang, Wanhao Liu, Shuchen Pu, Ben Gao, Jin Zeng, Xiaoyong Wei, Tianshu Yu, Shuzhou Sun, Tianfan Fu, Wanli Ouyang, Lei Bai, Jiatong Li, Zifu Wang, Yuqiang Li, Shufei Zhang

专题命中复杂问题求解：reasoning(abstract)

Comments 9 pages, 5 figures, 14 tables

2510.18327 2025-10-22 cs.SE 50%

InspectCoder: Dynamic Analysis-Enabled Self Repair through interactive LLM-Debugger Collaboration

Yunkun Wang, Yue Zhang, Guochang Li, Chen Zhi, Binhua Li, Fei Huang, Yongbin Li, Shuiguang Deng

专题命中复杂问题求解：reasoning(abstract)

2510.16610 2025-10-21 cs.CR 50%

Structuring Security: A Survey of Cybersecurity Ontologies, Semantic Log Processing, and LLMs Application

Bruno Lourenço, Pedro Adão, João F. Ferreira, Mario Monteiro Marques, Cátia Vaz

专题命中复杂问题求解：reasoning(abstract)

2510.13889 2025-10-17 cs.CV 50%

MultiFoodhat: A potential new paradigm for intelligent food quality inspection

Yue Hu, Guohang Zhuang

机构 * School of Food Science and Engineering, Central South University of Forestry and Technology（食品科学与工程学院，中南林业科技大学）； School of Computer and Information, Hefei University of Technology（计算机与信息学院，合肥工业大学）

专题命中复杂问题求解：reasoning(abstract)

2505.23179 2025-10-16 cs.CV 50%

DIP-R1: Deep Inspection and Perception with RL Looking Through and Understanding Complex Scenes

Sungjune Park, Hyunjun Kim, Junho Kim, Seongho Kim, Yong Man Ro

机构 * Integrated Vision and Language Lab., School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST)（整合视觉与语言实验室，电气工程学院，韩国科学技术院（KAIST））

专题命中复杂问题求解：reasoning(abstract)

2510.11907 2025-10-15 cs.CV 50%

Task-Specific Dual-Model Framework for Comprehensive Traffic Safety Video Description and Analysis

Blessing Agyei Kyem, Neema Jakisa Owor, Andrews Danyo, Joshua Kofi Asamoah, Eugene Denteh, Tanner Muturi, Anthony Dontoh, Yaw Adu-Gyamfi, Armstrong Aboah

机构 * North Dakota State University（北达科塔州立大学）； University of Missouri–Columbia（密苏里大学哥伦比亚分校）； University of Memphis（孟菲斯大学）

专题命中复杂问题求解：reasoning(abstract)

Comments This paper was accepted at ICCV 2025

2510.09963 2025-10-14 cs.RO 50%

LLM-HBT: Dynamic Behavior Tree Construction for Adaptive Coordination in Heterogeneous Robots

Chaoran Wang, Jingyuan Sun, Yanhui Zhang, Mingyu Zhang, Changju Wu

机构 * School of Aeronautic and Astronautics, Zhejiang University（浙江大学航空宇航学院）； Shanghai Huawei Technologies Co., Ltd.（上海华为技术有限公司）

专题命中复杂问题求解：reasoning(abstract)

Comments It contains 8 pages, 7 figures and 4 tables. This paper is submitted to ICRA 2026

1. 复杂问题求解 5035 篇

Backdooring CLIP through Concept Confusion

SafeGRPO: Self-Rewarded Multimodal Safety Alignment via Rule-Governed Policy Optimization

Automated Vulnerability Validation and Verification: A Large Language Model Approach

When AI Takes the Wheel: Security Analysis of Framework-Constrained Program Generation

EgoTV: Egocentric Task Verification from Natural Language Task Descriptions

UniVA: Universal Video Agent towards Open-Source Next-Generation Video Generalist

Empowering DINO Representations for Underwater Instance Segmentation via Aligner and Prompter

A Two-Stage System for Layout-Controlled Image Generation using Large Language Models and Diffusion Models

VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning

AesTest: Measuring Aesthetic Intelligence from Perception to Production

Gentle Manipulation Policy Learning via Demonstrations from VLM Planned Atomic Skills

Vector-Quantized Vision Foundation Models for Object-Centric Learning

ASAP: an Agentic Solution to Auto-optimize Performance of Large-Scale LLM Training

SVG Decomposition for Enhancing Large Multimodal Models Visualization Comprehension: A Study with Floor Plans

Pinpointing Trigger Moment for Grounded Video QA: Enhancing Spatio-temporal Grounding in Multimodal Large Language Models

Vote-in-Context: Turning VLMs into Zero-Shot Rank Fusers

PRevivor: Reviving Ancient Chinese Paintings using Prior-Guided Color Transformers

REaR: Retrieve, Expand and Refine for Effective Multitable Retrieval

Enhancing Temporal Understanding in Video-LLMs through Stacked Temporal Attention in Vision Encoders

PortGPT: Towards Automated Backporting Using Large Language Models

SeViCES: Unifying Semantic-Visual Evidence Consensus for Long Video Understanding

Beyond Text: Multimodal Jailbreaking of Vision-Language and Audio Models through Perceptually Simple Transformations

OmniVIC: A Self-Improving Variable Impedance Controller with Vision-Language In-Context Learning for Safe Robotic Manipulation

Chem-R: Learning to Reason as a Chemist

InspectCoder: Dynamic Analysis-Enabled Self Repair through interactive LLM-Debugger Collaboration

Structuring Security: A Survey of Cybersecurity Ontologies, Semantic Log Processing, and LLMs Application

MultiFoodhat: A potential new paradigm for intelligent food quality inspection

DIP-R1: Deep Inspection and Perception with RL Looking Through and Understanding Complex Scenes

Task-Specific Dual-Model Framework for Comprehensive Traffic Safety Video Description and Analysis

LLM-HBT: Dynamic Behavior Tree Construction for Adaptive Coordination in Heterogeneous Robots