AI Agent

DeepRNG: Towards Deep Reinforcement Learning-Assisted Generative Testing of Software

Chuan-Yung Tsai, Graham W. Taylor

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.LG、cs.SE

Comments Workshop on ML for Systems, 35th Conference on Neural Information Processing Systems (NeurIPS 2021)

ZERO: Playing Mathematical Programming Games

Gabriele Dragotto, Sriram Sankaranarayanan, Margarida Carvalho, Andrea Lodi

专题命中 软件智能体 ：agent(abstract)；multi-agent(abstract)

Explanation-Based Human Debugging of NLP Models: A Survey

Piyawat Lertvittayakumjorn, Francesca Toni

专题命中 软件智能体 ：workflow(abstract)；分类 cs.AI、cs.CL、cs.LG

Comments Accepted for publication at TACL. This version is a pre-MIT Press publication version

LOA: Logical Optimal Actions for Text-based Interaction Games

Daiki Kimura, Subhajit Chaudhury, Masaki Ono, Michiaki Tatsubori, Don Joven Agravante, Asim Munawar, Akifumi Wachi, Ryosuke Kohita, Alexander Gray

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.CL、cs.LG

Comments ACL-IJCNLP 2021 (demo paper)

Grounding Spatio-Temporal Language with Transformers

Tristan Karch, Laetitia Teodorescu, Katja Hofmann, Clément Moulin-Frier, Pierre-Yves Oudeyer

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.CL、cs.LG

Comments Contains main article and supplementaries

Journal ref Neurips 2021

IReEn: Reverse-Engineering of Black-Box Functions via Iterative Neural Program Synthesis

Hossein Hajipour, Mateusz Malinowski, Mario Fritz

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.LG、cs.SE

Comments 15 pages, 7 figures

Progressive Coordinate Transforms for Monocular 3D Object Detection

Li Wang, Li Zhang, Yi Zhu, Zhi Zhang, Tong He, Mu Li, Xiangyang Xue

专题命中 软件智能体 ：agent(abstract)；AI agent(abstract)

Comments Code is available at: https://github.com/amazon-research/progressive-coordinate-transforms

Detecting Anomalies in Semantic Segmentation with Prototypes

Dario Fontanel, Fabio Cermelli, Massimiliano Mancini, Barbara Caputo

专题命中 软件智能体 ：agent(abstract)；autonomous agent(abstract)

Comments SAIAD CVPR21 Workshop

A Joint Reinforcement-Learning Enabled Caching and Cross-Layer Network Code for Sum-Rate Maximization in F-RAN with D2D Communications

Mohammed S. Al-Abiad, Md. Zoheb Hassan, Md. Jahangir Hossain

专题命中 软件智能体 ：agent(abstract)；multi-agent(abstract)

Comments 15 pages, 9 figures, journal

Pathological Visual Question Answering

Xuehai He, Zhuo Cai, Wenlan Wei, Yichen Zhang, Luntian Mou, Eric Xing, Pengtao Xie

专题命中 软件智能体 ：agent(abstract)；AI agent(abstract)

Comments arXiv admin note: text overlap with arXiv:2003.10286

CPL-SLAM: Efficient and Certifiably Correct Planar Graph-Based SLAM Using the Complex Number Representation

Taosha Fan, Hanlin Wang, Michael Rubenstein, Todd Murphey

专题命中 软件智能体 ：agent(abstract)；autonomous agent(abstract)

Journal ref IEEE Transactions on Robotics, 2020

Attention Based Natural Language Grounding by Navigating Virtual Environment

Akilesh B, Abhishek Sinha, Mausoom Sarkar, Balaji Krishnamurthy

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.CL、cs.LG

Comments Accepted at WACV 2019. Also at NeurIPS 2017 workshop on Visually-Grounded Interaction and Language (ViGIL)

Question Asking as Program Generation

Anselm Rothe, Brenden M. Lake, Todd M. Gureckis

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.CL、cs.LG

Comments Published in Advances in Neural Information Processing Systems (NIPS) 30, December 2017

Journal ref Rothe, A., Lake, B. M., and Gureckis, T. M. (2017). Question asking as program generation. Advances in Neural Information Processing Systems 30

Global Stabilization of Triangulated Formations

Xudong Chen, M. -A. Belabbas, Tamer Başar

专题命中 软件智能体 ：agent(abstract)；autonomous agent(abstract)

Software Agents Interaction Algorithms in Virtual Learning Environment

Zahi A. M. Abu Sarhan

专题命中 软件智能体 ：agent(abstract)；multi-agent(abstract)

Journal ref The World of Computer Science and Information Technology Journal (WSCIT). 2014, Volume 4, Issue 2. pp. 18.25

MAS for video objects segmentation and tracking based on active contours and SURF descriptor

Mohamed Chakroun, Ali Wali, Adel M. Alimi

专题命中 软件智能体 ：agent(abstract)；multi-agent(abstract)

Comments 6 pages

Journal ref IJCSI International Journal of Computer Science Issues, Vol. 10, Issue 2, No 3, March 2013

Investigating the use of Software Agents to Reduce The Risk of Undetected Errors in Strategic Spreadsheet Applications

Pat Cleary, Dr David Ball, Mukul Madahar, Simon Thorne, Christopher Gosling, Karen Fernandez

专题命中 软件智能体 ：agent(abstract)；planning(abstract)

Comments 12 Pages, 3 Tables, 3 Figures

Journal ref Proc. European Spreadsheet Risks Int. Grp. (EuSpRIG) 2003 147-159 ISBN 1 86166 199 1

Cross-Model LLM Code Review: Should you use Claude to review Codex or vice versa?

跨模型大语言模型代码审查：应该用Claude审查Codex还是相反？

Zuodong Xiang, Yike Zhang, YueMing Zhang, Hailu Xu

机构 * University of California, Davis（加州大学戴维斯分校） ； Johns Hopkins University（约翰霍普金斯大学） ； California State University, Long Beach（长滩加州州立大学） 纠错

专题命中 软件智能体 ：workflow(abstract)；分类 cs.AI、cs.SE；agentic(comments)

AI总结 研究开发者同时使用Claude和Codex进行代码审查的成本、时间及配对顺序问题，通过对116个任务的六种条件实验发现，Claude审查Codex草稿效果好，反向则不佳，有用的配对是不对称的，应Claude审查Codex。

Comments This paper had been accepted by Agentic SE @ KDD'26

Modelling Customer Trajectories with Reinforcement Learning for Practical Retail Insights

用强化学习建模客户轨迹以获得实际零售洞察

Ken Ming Lee, Paul Barde, Maxime C. Cohen, Derek Nowrouzezahrai

机构 * McGill University（麦吉尔大学） ； Mila - Quebec AI Institute（魁北克人工智能研究所） 纠错

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.LG；autonomous agent(comments)

AI总结 本文提出了一种基于智能体的建模框架，将客户轨迹预测转化为最大熵强化学习问题，以更准确地反映具有有限理性的客户行为，从而提供更精确的冲动购买率和货架交通密度估计。

Comments Proceeding of the 25th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2026)

Banana100: Breaking NR-IQA Metrics by 100 Iterative Image Replications with Nano Banana Pro

Banana100: 通过100次迭代图像复制打破NR-IQA度量标准

Kenan Tang, Praveen Arunshankar, Andong Hua, Anthony Yang, Yao Qin

机构 * University of California, Santa Barbara（加州大学圣塔芭芭拉分校） 纠错

专题命中 软件智能体 ：agentic(abstract,comments)；分类 cs.AI、cs.LG

AI总结 Banana100通过100次迭代编辑生成28000张退化图像，揭示多轮编辑中图像质量退化问题，发现现有NR-IQA度量标准无法检测严重退化图像，威胁未来模型训练稳定性。

Comments Accepted to CVPR 2026 Workshop on Agentic AI for Visual Media

Reflection-Driven Control for Trustworthy Code Agents

基于反射的可信代码代理控制

Bin Wang, Jiazheng Quan, Xingrui Yu, Hansen Hu, Yuhao, Ivor Tsang

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.SE；agentic(comments)

AI总结 本文提出反射驱动控制方法，通过内部反思循环提升代码生成的安全性和合规性，实现自主、安全且可审计的AI代码代理。

Comments Accepted to AAAI 2026 Workshop on Trust and Control in Agentic AI (TrustAgent)

Text-to-SQL for Enterprise Data Analytics

Albert Chen, Manas Bundele, Gaurav Ahlawat, Patrick Stetz, Zhitao Wang, Qiang Fei, Donghoon Jung, Audrey Chu, Bharadwaj Jayaraman, Ayushi Panth, Yatin Arora, Sourav Jain, Renjith Varma, Alexey Ilin, Iuliia Melnychuk, Chelsea Chueh, Joyan Sil, Xiaofeng Wang

机构 * LinkedIn（领英） 纠错

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.CL；agentic(comments)

Comments 11 pages, 8 figures, Workshop on Agentic AI for Enterprise at KDD '25

SOPBench: Evaluating Language Agents at Following Standard Operating Procedures and Constraints

Zekun Li, Shinda Huang, Jiangtian Wang, Nathan Zhang, Antonis Antoniades, Wenyue Hua, Kaijie Zhu, Sirui Zeng, Chi Wang, William Yang Wang, Xifeng Yan

专题命中 软件智能体 ：agent(abstract,comments)；分类 cs.AI、cs.CL

Comments Code, data, and over 24k agent trajectories are released at https://github.com/Leezekun/SOPBench

A Grounded Interaction Protocol for Explainable Artificial Intelligence

Prashan Madumal, Tim Miller, Liz Sonenberg, Frank Vetere

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.LG；autonomous agent(comments)

Comments To appear in 18th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2019) as a full paper. arXiv admin note: substantial text overlap with arXiv:1806.08055

Beyond Satisfaction: From Placebic to Actionable Explanations For Enhanced Understandability

超越满意：从安慰性到可操作性解释以提升可理解性

Joe Shymanski, Jacob Brue, Sandip Sen

机构 * The University of Tulsa（图兰大学） 纠错

专题命中 软件智能体 ：agent(abstract,journal_ref)；分类 cs.AI；multi-agent(journal_ref)

AI总结 本文探讨了可解释性在提升系统可理解性中的作用，通过实验发现可操作性解释在任务表现上优于安慰性解释，但用户满意度评分相同，强调需结合客观指标与主观评估来衡量解释质量。

Comments 21 pages, 7 figures, 6 tables. EXTRAAMAS 2025 submission. Preprint version

Journal ref In: Calvaresi, D., et al. Explainable, Trustworthy, and Responsible AI and Multi-Agent Systems. EXTRAAMAS 2025. Lecture Notes in Computer Science. Springer, Cham

Towards an LLM-powered Social Digital Twinning Platform

Önder Gürcan, Vanja Falck, Markus G. Rousseau, Larissa L. Lima

机构 * Center for Modeling Social Systems（社会科学建模中心） ； NORCE Norwegian Research Center AS（挪威NORCE研究机构） ； Kristiansand, Norway（挪威克里斯蒂安桑） 纠错

专题命中 软件智能体 ：agent(abstract,comments)；分类 cs.AI；multi-agent(comments)

Comments 13 pages, 3 figures, 23rd International Conference on Practical applications of Agents and Multi-Agent Systems (PAAMS 2025)

Disagree to Explore, Agree to Commit: Routing-Guided Test-Time Scaling for Software Agents

探索时持异议，提交时持共识：面向软件智能体的路由引导测试时缩放

Kang Chen, Junjie Nian, Yixin Cao, Yugang Jiang

机构 * Fudan University（复旦大学） ； Shanghai Innovation Institute（上海创新研究院） 纠错

专题命中 软件智能体 ：tool-use(abstract)；分类 cs.AI、cs.SE

AI总结 该研究针对软件智能体测试时缩放难题，提出Risa方法，利用MoE路由轨迹引导探索与仲裁，在SWE-bench等基准上提升了仓库级软件工程任务的解决率。

Repo2Skill-Evo: Repository Skills Go Stale in Silence

Repo2Skill-Evo：仓库技能在静默中过时

Chenyuan Duan, Ge Shi, Zineng Mao, Ge Zhang, Hao Liang, Yinzhu Piao, Yuchen Wu, Zhixin Yao, Kaiyu Huang, Wenhao Huang, Linzhuang Sun, Shen Yan, Wentao Zhang

机构 * ByteDance（字节跳动） ； Peking University（北京大学） ； Beijing Jiaotong University（北京交通大学） 纠错

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.SE

AI总结 Repo2Skill-Evo研究发现，在57个真实仓库的105次版本转换中，前沿智能体难以可靠维护仓库技能，其平均@3 macro F1仅29.9%-69.7%，仓库技能会在无明确信号的情况下静默过时。

GameXpert-Bench: How Far Are Coding Agents from Expert Game Development?

GameXpert-Bench：编码智能体距离专家级游戏开发还有多远？

Kun Chen, Haorong Hong, Peizhong Gao, Jianfeng Lin, Tongxu Luo, Yuxuan Xie, Chenxu Liu, Jieling He, Zhongyuan Liu, Zeno Zeng

机构 * Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所） ； Tsinghua University（清华大学） ； The Hong Kong University of Science and Technology（香港科技大学） ； The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳）） ； Lightspeed Studios, Tencent（腾讯光速工作室） 纠错

专题命中 软件智能体 ：agent(abstract)；分类 cs.AI、cs.CL

AI总结 GameXpert-Bench是覆盖游戏开发全生命周期的基准，含三个赛道，测评显示当前编码智能体在生成可玩游戏基础上表现较好，在缺陷发现等方面仍有不足。

AI with Authority, from Application to Silicon

具备权威的AI：从应用到硅片

Jason Hickey

专题命中 软件智能体 ：AI agent(abstract)；分类 cs.AI、cs.SE

AI总结 该研究提出Salt方法，借助生成式AI与机器验证，在五周内由一名研究人员指挥AI智能体完成RISC-V处理器的流片，无人工审核证明与RTL，实现高效可靠的自主机器工作。

Comments 17 pages, 6 figures