AI Agent

Behavioral Cloning via Search in Video PreTraining Latent Space

Federico Malato, Florian Leopold, Amogh Raut, Ville Hautamäki, Andrew Melnik

专题命中 工具调用 ：agent(abstract)；autonomous agent(abstract)；分类 cs.AI、cs.LG

Monte Carlo Tree Search Algorithms for Risk-Aware and Multi-Objective Reinforcement Learning

Conor F. Hayes, Mathieu Reymond, Diederik M. Roijers, Enda Howley, Patrick Mannion

专题命中 工具调用 ：agent(abstract)；planning(abstract)；分类 cs.AI、cs.LG

Comments arXiv admin note: substantial text overlap with arXiv:2102.00966

Learning to design without prior data: Discovering generalizable design strategies using deep learning and tree search

Ayush Raina, Jonathan Cagan, Christopher McComb

专题命中 工具调用 ：agent(abstract)；AI agent(abstract)；分类 cs.AI、cs.LG

Comments ASME. J. Mech. Des

AlphaZero-Inspired Game Learning: Faster Training by Using MCTS Only at Test Time

Johannes Scheiermann, Wolfgang Konen

专题命中 工具调用 ：agent(abstract)；planning(abstract)；分类 cs.AI、cs.LG

Comments 11 pages, 10 figures

Chain of Thought Imitation with Procedure Cloning

Mengjiao Yang, Dale Schuurmans, Pieter Abbeel, Ofir Nachum

专题命中 工具调用 ：agent(abstract)；planning(abstract)；分类 cs.AI、cs.LG

Modeling Strong and Human-Like Gameplay with KL-Regularized Search

Athul Paul Jacob, David J. Wu, Gabriele Farina, Adam Lerer, Hengyuan Hu, Anton Bakhtin, Jacob Andreas, Noam Brown

专题命中 工具调用 ：agent(abstract)；multi-agent(abstract)；分类 cs.AI、cs.LG

Pretrained Cost Model for Distributed Constraint Optimization Problems

Yanchen Deng, Shufeng Kong, Bo An

专题命中 工具调用 ：agent(abstract)；autonomous agent(abstract)；分类 cs.AI、cs.LG

Comments Accepted by AAAI-22

Counter-Strike Deathmatch with Large-Scale Behavioural Cloning

Tim Pearce, Jun Zhu

专题命中 工具调用 ：agent(abstract)；AI agent(abstract)；分类 cs.AI、cs.LG

Comments Offline Reinforcement Learning Workshop at Neural Information Processing Systems, 2021

Training Agents using Upside-Down Reinforcement Learning

Rupesh Kumar Srivastava, Pranav Shyam, Filipe Mutz, Wojciech Jaśkowski, Jürgen Schmidhuber

专题命中 工具调用 ：agent(abstract)；autonomous agent(abstract)；分类 cs.AI、cs.LG

Comments Extends NeurIPS 2019 Deep Reinforcement Learning workshop presentation

Dark, Beyond Deep: A Paradigm Shift to Cognitive AI with Humanlike Common Sense

Yixin Zhu, Tao Gao, Lifeng Fan, Siyuan Huang, Mark Edmonds, Hangxin Liu, Feng Gao, Chi Zhang, Siyuan Qi, Ying Nian Wu, Joshua B. Tenenbaum, Song-Chun Zhu

专题命中 工具调用 ：tool use(abstract)；planning(abstract)；分类 cs.AI、cs.LG

Comments For high quality figures, please refer to http://wellyzhang.github.io/attach/dark.pdf

Journal ref Engineering, Feb, 2020

On Solving Cooperative MARL Problems with a Few Good Experiences

Rajiv Ranjan Kumar, Pradeep Varakantham

专题命中 工具调用 ：agent(abstract)；multi-agent(abstract)；分类 cs.AI、cs.LG

What can you do with a rock? Affordance extraction via word embeddings

Nancy Fulda, Daniel Ricks, Ben Murdoch, David Wingate

专题命中 工具调用 ：agent(abstract)；autonomous agent(abstract)；分类 cs.AI、cs.CL

Comments 7 pages, 7 figures, 2 algorithms, data runs were performed using the Autoplay learning environment for interactive fiction

Journal ref Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence (IJCAI), Pages 1039-1045, 2017

Imitation from Observation: Learning to Imitate Behaviors from Raw Video via Context Translation

YuXuan Liu, Abhishek Gupta, Pieter Abbeel, Sergey Levine

专题命中 工具调用 ：agent(abstract)；tool use(abstract)；分类 cs.AI、cs.LG

Comments Accepted at ICRA 2018, Brisbane. YuXuan Liu and Abhishek Gupta had equal contribution

Feedback-Based Tree Search for Reinforcement Learning

Daniel R. Jiang, Emmanuel Ekwedike, Han Liu

专题命中 工具调用 ：agent(abstract)；AI agent(abstract)；分类 cs.AI、cs.LG

Comments 19 pages, to be presented at ICML 2018

Learning at the Ends: From Hand to Tool Affordances in Humanoid Robots

Giovanni Saponaro, Pedro Vicente, Atabak Dehban, Lorenzo Jamone, Alexandre Bernardino, José Santos-Victor

专题命中 工具调用 ：tool use(abstract)；planning(abstract)；分类 cs.AI、cs.LG

Comments dataset available at htts://vislab.isr.tecnico.ulisboa.pt/, IEEE International Conference on Development and Learning and on Epigenetic Robotics (ICDL-EpiRob 2017)

DialPort: Connecting the Spoken Dialog Research Community to Real User Data

Tiancheng Zhao, Kyusong Lee, Maxine Eskenazi

专题命中 工具调用 ：agent(abstract)；multi-agent(abstract)；分类 cs.AI、cs.CL

Comments Under Peer Review of SigDial 2016

Learning for Adaptive Real-time Search

Vadim Bulitko

专题命中 工具调用 ：autonomous agent(abstract)；planning(abstract)；分类 cs.AI、cs.LG

Lomekwi: Resource-Bounded Tool Discovery in LLM Agents

洛梅奎：大语言模型智能体中资源受限的工具发现

Roshan Klein-Seetharaman, Daniel Wang, Andrew Xu

机构 * Sea12 Technologies（Sea12科技公司） ； Yale University（耶鲁大学） 纠错

专题命中 工具调用 ：tool use(abstract)；tool-use(abstract)；分类 cs.AI；agentic(comments)

AI总结 研究受认知科学启发区分工具使用与发现，将工具发现分解为好奇心、识别和效率，表明该框架可用于现有任务，证明识别与模型大小成反比，还通过组合博弈及模拟环境观察到反比缩放。

Comments All authors contributed equally. 17 pages, 6 figures. Presented at the 2026 Conference on Learning Theory Workshop on Learning in an Agentic World

optimize_anything: A Universal API for Optimizing any Text Parameter

optimize_anything: 一个用于优化任何文本参数的通用API

Lakshya A Agrawal, Donghyun Lee, Shangyin Tan, Wenjie Ma, Karim Elmaaroufi, Rohit Sandadi, Sanjit A. Seshia, Koushik Sen, Dan Klein, Ion Stoica, Joseph E. Gonzalez, Omar Khattab, Alexandros G. Dimakis, Matei Zaharia

机构 * MIT（麻省理工学院） 纠错

专题命中 工具调用 ：agent(abstract)；分类 cs.AI、cs.CL、cs.LG；agentic(journal_ref)

AI总结 本文提出了一种基于LLM的通用优化系统，能够跨不同领域实现文本参数的优化，展示了其在六个多样化任务中的state-of-the-art性能，通过多任务搜索和跨问题迁移实现了高效的优化。

Comments 16 pages, 11 figures; Blog: https://gepa-ai.github.io/gepa/blog/2026/02/18/introducing-optimize-anything/

Journal ref Proceedings of the ACM Conference on AI and Agentic Systems (CAIS 26), May 26-29, 2026, San Jose, CA, USA

Information Fidelity in Tool-Using LLM Agents: A Martingale Analysis of the Model Context Protocol

工具使用LLM代理中的信息保真度：模型上下文协议的鞅分析

Flint Xiaofeng Fan, Cheston Tan, Roger Wattenhofer, Yew-Soon Ong

专题命中 工具调用 ：agent(abstract)；AI agent(abstract)；分类 cs.AI；autonomous agent(comments)

AI总结 本文提出了一种分析模型上下文协议中误差累积的理论框架，通过鞅分析证明误差呈现线性增长并受$O(\sqrt{T})$约束，实验验证了语义加权和周期性再定位对误差控制的有效性。

Comments Full working version of an extended abstract accepted at the 25th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2026)

Learning From Failure: Integrating Negative Examples when Fine-tuning Large Language Models as Agents

Renxi Wang, Haonan Li, Xudong Han, Yixuan Zhang, Timothy Baldwin

专题命中 工具调用 ：agent(abstract,comments)；tool use(abstract)；分类 cs.CL

Comments Agent, LLM, Large Language Model

Improving Search by Utilizing State Information in OPTIC Planners Compilation to LP

Elad Denenberg, Amanda Coles, Derek Long

专题命中 工具调用 ：agent(abstract)；autonomous agent(abstract)；分类 cs.AI；planning(comments)

Comments 8 pages, 3 figures. Preprint, last submitted to the International Conference on Automated Planning and Scheduling (ICAPS 2021) at 21.01.2021

Effective Game-Theoretic Motion Planning via Nested Search

通过嵌套搜索实现有效的博弈论运动规划

Avishav Engle, Andrey Zhitnikov, Oren Salzman, Omer Ben-Porat, Kiril Solovey

机构 * Technion–Israel Institute of Technology（技术离子以色列理工学院） 纠错

专题命中 工具调用 ：planning(title)

AI总结 本文提出GTNS方法，通过嵌套搜索有效计算博弈论中的纳什均衡，适用于自动驾驶等场景，实现快速且可靠的运动规划。

Comments Updated version. Offline graph creation runtime added. Acknowledgements added

GSpyNetTree-O4: an event validation tool used in the fourth LIGO-Virgo-KAGRA observing run

GSpyNetTree - O4：第四次LIGO - Virgo - KAGRA观测运行中使用的事件验证工具

Sofia Alvarez-Lopez, Man Leong Chan, Franz S. Herbst, Dhatri Raghunathan, Airene Ahuja, Annudesh Liyanage, Julian Ding, Alejandro Garcia-Varela, Raymond Ng, Jess McIver

专题命中 工具调用 ：tool use(title)

AI总结 研究针对引力波探测器数据中的毛刺问题，在第四次LIGO - Virgo - KAGRA观测运行中部署GSpyNetTree - O4工具。通过新架构、扩充训练集及校准校正等方法，该工具在毛刺识别和无毛刺样本判断上表现良好，提高了引力波事件验证工作流程的自动化。

Comments 23 pages, 16 figures

Splitting method for spatio-temporal search efforts planning

时空搜索努力规划的拆分方法

Chouchane Mathieu, Paris Sébastien, Le Gland François, Ouladsine Mustapha

专题命中 工具调用 ：planning(title)

AI总结 提出基于广义拆分方法的新型随机优化算法，用于最大化检测智能随机移动目标的概率，无需状态空间离散化且能处理多种约束。

Comments Article rejected

Meta-Adaptive Beam Search Planning for Transformer-Based Reinforcement Learning Control of UAVs with Overhead Manipulators under Flight Disturbances

基于Transformer强化学习的无人机搭载顶置机械臂在飞行扰动下的元自适应波束搜索规划

Hazim Alzorgan, Sayed Pedram Haeri Boroujeni, Abolfazl Razi

专题命中 工具调用 ：planning(title)

AI总结 针对无人机与顶置机械臂耦合导致的末端执行器跟踪误差问题，提出基于Transformer双深度Q网络（DDQN）的强化学习框架，通过自适应波束搜索规划器利用学习到的评论家进行前向估计，实现软件在环的短视域波束搜索，显著降低跟踪误差并提升奖励。

Comments The paper will be reworked significantly

AgentSecBench: Measuring Prompt Injection, Privacy Leakage, and Tool-Use Integrity in LLM Agents

AgentSecBench：测量LLM智能体中的提示注入、隐私泄露和工具使用完整性

Faruk Alpay, Taylan Alpay

专题命中 工具调用 ：tool-use(title)

AI总结 提出AgentSecBench框架，通过定义意图到执行的无干扰游戏，系统评估LLM智能体在指令完整性、检索机密性和能力完整性方面的安全风险，并实验验证防御措施的有效性。

Comments 24 pages, 3 figures. Ancillary files provided

Patho-AgenticRAG: Towards Multimodal Agentic Retrieval-Augmented Generation for Pathology VLMs via Reinforcement Learning

病理代理RAG：通过强化学习实现多模态代理检索增强生成用于病理学视觉语言模型

Wenchuan Zhang, Jingru Guo, Hengzhe Zhang, Penghao Zhang, Jie Chen, Shuwan Zhang, Zhang Zhang, Yuhao Yi, Hong Bu

专题命中 工具调用 ：agentic(title)

AI总结 本文提出Patho-AgenticRAG，通过强化学习实现多模态代理检索增强生成，解决病理学视觉语言模型在高分辨率、复杂组织结构和临床语义上的挑战，提升诊断准确性。

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence, 40(35): 29921-29929, 2026

AeroGen: Agentic Drone Autonomy through Single-Shot Structured Prompting & Drone SDK

AeroGen：通过单次结构提示与无人机SDK实现代理无人机自主性

Kautuk Astu, Yogesh Simmhan

专题命中 工具调用 ：agentic(title)

AI总结 本文提出AeroGen框架，通过结构化提示和AeroDaaS SDK实现无人机自主控制程序的可靠生成，验证了其在多种环境下的有效性。

RPT*: Global Planning with Probabilistic Terminals for Target Search in Complex Environments

RPT*: 带概率终端的全局规划用于复杂环境中的目标搜索

Yunpeng Lyu, Chao Cao, Ji Zhang, Howie Choset, Zhongqiang Ren

机构 * Global College, Shanghai Jiao Tong University（上海交通大学全球学院） ； Robotics Institute, Carnegie Mellon University（卡内基梅隆大学机器人研究所） 纠错

专题命中 工具调用 ：planning(title)

AI总结 RPT*通过动态规划和新启发式方法解决复杂环境中的目标搜索问题，实现最优路径规划与高效探索的平衡。