AI Open最新文献

英文中文

IF 14.8

AI Open

Pub Date : 2025-01-01

引用次数: 0

ADAP: Adaptive & Dynamic Arc Padding for predicting seam profiles in Multi-Layer-Multi-Pass robotic welding ADAP：用于多层多道次机器人焊接焊缝轮廓预测的自适应动态电弧填充

IF 14.8

AI Open

Pub Date : 2025-01-01 DOI: 10.1016/j.aiopen.2025.10.003

He Wang , Sen Li , Xiaobo Liu , Chengxiao Dong , Fang Wan

Welding thick metal plates using Multi-Layer-Multi-Pass (MLMP) techniques demands precise control over the weld seam profile as it evolves during the cooling process. In MLMP welding, typically executed with Gas Metal Arc Welding (GMAW) and shielding gas protection, the continuous deposition of weld beads results in dynamic changes to the seam geometry. These challenging traditional robotic welding systems rely on static models. To ensure high-quality joints, real-time adaptation of welding paths requires accurate predictions of weld bead geometry, which in turn guide the estimation of welding positions for adaptive trajectory planning. In this study, we introduce the Adaptive & Dynamic Arc Padding (ADAP) framework. This novel data-driven approach integrates deep learning with an innovative arc-based representation of weld bead profiles. By representing the weld bead geometry through image-derived boundaries and primitive arc parameters (arc center and radius), ADAP establishes a direct link between welding parameters and the evolving weld seam profile. Utilizing datasets generated from Flow-3D simulations of the MLMP process, our framework achieves high-accuracy, real-time predictions: welding positions are estimated within 0.025 s (with an average error of approximately 1.5 mm), and weld seam profiles are predicted in 15 ms, with the arc-based geometric parameters accurately estimated (average errors of 0.73 mm in arc center position and 0.66 mm in radius). This practical approach enhances the efficiency and quality of MLMP robotic welding and contributes to advances in data-driven modeling and intelligent control in manufacturing, paving the way for autonomous welding systems.

采用多层多道次（MLMP）技术焊接厚金属板需要精确控制焊缝轮廓，因为它在冷却过程中的演变。在MLMP焊接中，通常采用气体金属弧焊（GMAW）和保护气体保护，焊接珠的连续沉积导致焊缝几何形状的动态变化。这些具有挑战性的传统机器人焊接系统依赖于静态模型。为了确保高质量的接头，焊接路径的实时自适应需要准确预测焊缝几何形状，从而指导自适应轨迹规划中焊接位置的估计。在本研究中，我们介绍了自适应动态弧线填充（ADAP）框架。这种新颖的数据驱动方法将深度学习与基于弧线的焊缝轮廓表示相结合。通过图像派生的边界和原始电弧参数（电弧中心和半径）表示焊缝几何形状，ADAP在焊接参数和不断变化的焊缝轮廓之间建立了直接联系。利用MLMP过程Flow-3D模拟生成的数据集，我们的框架实现了高精度的实时预测：焊接位置在0.025 s内估计（平均误差约为1.5 mm），焊缝轮廓在15 ms内预测，基于电弧的几何参数准确估计（电弧中心位置平均误差为0.73 mm，半径平均误差为0.66 mm）。这种实用的方法提高了MLMP机器人焊接的效率和质量，有助于在制造业中数据驱动建模和智能控制的进步，为自主焊接系统铺平了道路。

{"title":"ADAP: Adaptive & Dynamic Arc Padding for predicting seam profiles in Multi-Layer-Multi-Pass robotic welding","authors":"He Wang , Sen Li , Xiaobo Liu , Chengxiao Dong , Fang Wan","doi":"10.1016/j.aiopen.2025.10.003","DOIUrl":"10.1016/j.aiopen.2025.10.003","url":null,"abstract":"<div><div>Welding thick metal plates using Multi-Layer-Multi-Pass (MLMP) techniques demands precise control over the weld seam profile as it evolves during the cooling process. In MLMP welding, typically executed with Gas Metal Arc Welding (GMAW) and shielding gas protection, the continuous deposition of weld beads results in dynamic changes to the seam geometry. These challenging traditional robotic welding systems rely on static models. To ensure high-quality joints, real-time adaptation of welding paths requires accurate predictions of weld bead geometry, which in turn guide the estimation of welding positions for adaptive trajectory planning. In this study, we introduce the Adaptive & Dynamic Arc Padding (ADAP) framework. This novel data-driven approach integrates deep learning with an innovative arc-based representation of weld bead profiles. By representing the weld bead geometry through image-derived boundaries and primitive arc parameters (arc center and radius), ADAP establishes a direct link between welding parameters and the evolving weld seam profile. Utilizing datasets generated from Flow-3D simulations of the MLMP process, our framework achieves high-accuracy, real-time predictions: welding positions are estimated within 0.025 s (with an average error of approximately 1.5 mm), and weld seam profiles are predicted in 15 ms, with the arc-based geometric parameters accurately estimated (average errors of 0.73 mm in arc center position and 0.66 mm in radius). This practical approach enhances the efficiency and quality of MLMP robotic welding and contributes to advances in data-driven modeling and intelligent control in manufacturing, paving the way for autonomous welding systems.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"6 ","pages":"Pages 204-219"},"PeriodicalIF":14.8,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145361833","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

LLMKG＋: Systematically improving knowledge quality and coverage in KGs using LLMs – A case study in medical domain LLMKG+：使用LLMs系统地提高知识质量和覆盖范围-医学领域的案例研究

IF 14.8

AI Open

Pub Date : 2025-01-01 DOI: 10.1016/j.aiopen.2025.11.003

Xincan Feng , Hejie Cui , Kazuki Hayashi , Huy Hien Vu , Kenta T. Suzuki , Noriki Nishida , Hidetaka Kamigaito , Yuji Matsumoto , Taro Watanabe , Carl Yang

Knowledge graphs (KGs) encode structured information about real-world entities and their relations, supporting core NLP tasks such as question answering and retrieval. Existing LLM-based methods for knowledge extraction and fusion often struggle to balance quality and coverage when adapting to emerging knowledge. We propose LLMKG＋, a framework for KG expansion that integrates the generative strengths of LLMs with relevance verification. LLMKG＋features (1) a two-stage pipeline with retrieval-augmented generation followed by hierarchical expansion filtering, where the latter is the first to jointly assess semantic equivalence to eliminate triple-level redundancy while ensuring factual correctness, and (2) a novel KG Reconstruction Test that recognizes semantically equivalent triples to enable more accurate quality and coverage assessment. Evaluated on PubMed abstracts and the UMLS semantic network using eight state-of-the-art LLMs, LLMKG＋improves KG quality and coverage by 20.47%–73.71% over strong baselines. These results demonstrate that LLMKG＋offers an effective solution for KG expansion in domains requiring high quality, broad coverage, and continual knowledge growth. Code: https://github.com/xincanfeng/llmkg.

知识图（Knowledge graphs, KGs）对现实世界实体及其关系的结构化信息进行编码，支持诸如问题回答和检索等核心NLP任务。现有的基于法学硕士的知识提取和融合方法在适应新兴知识时往往难以平衡质量和覆盖范围。我们提出LLMKG+，这是一个将llm的生成优势与相关性验证相结合的KG扩展框架。LLMKG+具有以下特点：(1)检索增强生成和分层扩展过滤的两阶段管道，其中分层扩展过滤首先联合评估语义等价性，以消除三级冗余，同时确保事实正确性；(2)识别语义等价三重的新颖KG重构测试，以实现更准确的质量和覆盖评估。使用8个最先进的llm对PubMed摘要和UMLS语义网络进行评估，LLMKG+在强基线上提高了KG的质量和覆盖率20.47%-73.71%。这些结果表明，LLMKG+为需要高质量、广泛覆盖和持续知识增长的领域的KG扩展提供了有效的解决方案。代码:https://github.com/xincanfeng/llmkg。

{"title":"LLMKG＋: Systematically improving knowledge quality and coverage in KGs using LLMs – A case study in medical domain","authors":"Xincan Feng , Hejie Cui , Kazuki Hayashi , Huy Hien Vu , Kenta T. Suzuki , Noriki Nishida , Hidetaka Kamigaito , Yuji Matsumoto , Taro Watanabe , Carl Yang","doi":"10.1016/j.aiopen.2025.11.003","DOIUrl":"10.1016/j.aiopen.2025.11.003","url":null,"abstract":"<div><div>Knowledge graphs (KGs) encode structured information about real-world entities and their relations, supporting core NLP tasks such as question answering and retrieval. Existing LLM-based methods for knowledge extraction and fusion often struggle to balance quality and coverage when adapting to emerging knowledge. We propose LLMKG＋, a framework for KG expansion that integrates the generative strengths of LLMs with relevance verification. LLMKG＋features (1) a two-stage pipeline with retrieval-augmented generation followed by hierarchical expansion filtering, where the latter is the first to jointly assess semantic equivalence to eliminate triple-level redundancy while ensuring factual correctness, and (2) a novel KG Reconstruction Test that recognizes semantically equivalent triples to enable more accurate quality and coverage assessment. Evaluated on PubMed abstracts and the UMLS semantic network using eight state-of-the-art LLMs, LLMKG＋improves KG quality and coverage by 20.47%–73.71% over strong baselines. These results demonstrate that LLMKG＋offers an effective solution for KG expansion in domains requiring high quality, broad coverage, and continual knowledge growth. Code: <span><span>https://github.com/xincanfeng/llmkg</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"6 ","pages":"Pages 299-313"},"PeriodicalIF":14.8,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145623234","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Advancing AI for science: From the revolution of tools to the tools for revolution 为科学推进人工智能：从工具革命到工具革命

IF 14.8

AI Open

Pub Date : 2025-01-01 DOI: 10.1016/j.aiopen.2025.11.002

Bowen Zhou , Ning Ding , Lei Bai , Hao Zhou

Scientific research is not a linear pipeline but a dynamic system built upon the ever-shifting interactions among three elements — research objects, tools, and researchers. And sustained progress depends on how quickly insights circulate within this network, not on optimizing a single node in isolation. With the impending arrival of more general artificial intelligence, we stand at a critical point in how AI might change scientific research in a systemic manner. Recent “AI for Science” achievements – from protein-structure prediction to accelerated climate simulations – have proven the value of task-level AI-driven solutions. Yet, potential still remains unrealized when these advances are siloed in disciplinary “archipelagos”. This paper argues that the real prize is systemic: AI that simultaneously expands the research objects’ data landscape (AI for Data), rewires computational research tools (AI for Computation), and co-creates hypotheses with researchers (AI for Innovation). When these three pushes converge, AI stops being merely a revolution of tools but becomes the tool of revolution — a catalyst that raises the frequency, breadth, and depth of discovery across disciplines. By enhancing the full research triad rather than isolated nodes, AI can raise the overall tempo and scope of discovery in a measured, discipline-agnostic way.

科学研究不是一个线性的管道，而是一个动态的系统，它建立在研究对象、工具和研究人员这三个要素之间不断变化的相互作用之上。持续的进步取决于洞察力在这个网络中传播的速度，而不是孤立地优化单个节点。随着更通用的人工智能的到来，我们正处于人工智能如何系统性地改变科学研究的关键时刻。最近的“科学人工智能”成就——从蛋白质结构预测到加速气候模拟——已经证明了任务级人工智能驱动解决方案的价值。然而，当这些进步被孤立在学科“群岛”中时，潜力仍然没有实现。本文认为，真正的奖励是系统性的：人工智能同时扩展了研究对象的数据景观（数据人工智能），重新连接了计算研究工具（计算人工智能），并与研究人员共同创造了假设（创新人工智能）。当这三种推动力汇聚在一起时，人工智能就不再仅仅是一种工具革命，而是成为革命的工具——一种提高跨学科发现频率、广度和深度的催化剂。通过增强完整的研究三位一体，而不是孤立的节点，人工智能可以以一种有分寸的、学科不可知论的方式提高发现的整体速度和范围。

{"title":"Advancing AI for science: From the revolution of tools to the tools for revolution","authors":"Bowen Zhou , Ning Ding , Lei Bai , Hao Zhou","doi":"10.1016/j.aiopen.2025.11.002","DOIUrl":"10.1016/j.aiopen.2025.11.002","url":null,"abstract":"<div><div>Scientific research is not a linear pipeline but a dynamic system built upon the ever-shifting interactions among three elements — <em>research objects, tools, and researchers</em>. And sustained progress depends on how quickly insights circulate within this network, not on optimizing a single node in isolation. With the impending arrival of more general artificial intelligence, we stand at a critical point in how AI might change scientific research in a systemic manner. Recent “AI for Science” achievements – from protein-structure prediction to accelerated climate simulations – have proven the value of task-level AI-driven solutions. Yet, potential still remains unrealized when these advances are siloed in disciplinary “archipelagos”. This paper argues that the real prize is systemic: AI that simultaneously expands the research objects’ data landscape (AI for Data), rewires computational research tools (AI for Computation), and co-creates hypotheses with researchers (AI for Innovation). When these three pushes converge, AI stops being merely a revolution of tools but becomes the tool of revolution — a catalyst that raises the frequency, breadth, and depth of discovery across disciplines. By enhancing the full research triad rather than isolated nodes, AI can raise the overall tempo and scope of discovery in a measured, discipline-agnostic way.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"6 ","pages":"Pages 323-328"},"PeriodicalIF":14.8,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145692937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Optimal RoPE extension via Bayesian Optimization for training-free length generalization 基于贝叶斯优化的无训练长度泛化的最优RoPE扩展

AI Open

Pub Date : 2025-01-01 DOI: 10.1016/j.aiopen.2025.01.002

Xinrong Zhang , Shengding Hu , Weilin Zhao , Huadong Wang , Xu Han , Chaoqun He , Guoyang Zeng , Zhiyuan Liu , Maosong Sun

Transformers are designed to process input of variable length without resource constraints. However, their performance significantly deteriorates when the input surpasses a threshold slightly larger than the pre-training context window. This limitation on the effective context window confines the application of Transformer-based large language models (LLMs) that have been the subject of great anticipation. Consequently, the generalization of pre-trained LLMs to handle varying input lengths becomes a pivotal and formidable challenge. Previous research has endeavored to address this challenge by modifying the Rotary Position Embedding (RoPE), the primary factor responsible for disparities in handling different input lengths. These efforts have provided valuable insights, while they often lack a deep understanding of the root causes of performance degradation and rely heavily on manual parameter tuning. In response to these issues, we conduct a comprehensive analysis and identify two primary causes behind the performance drop: global distribution mismatch and local resolution degradation. In light of these challenges, we introduce an Optimal RoPE (ORoPE) extension using Bayesian Optimization (BO), which alleviates the need for additional model training. Our experiments demonstrate the efficacy of our approach, outperforming baselines by up to 21.9%, 32.1%, and 41.2% at evaluation lengths of 8K, 16K, and 32K, respectively. We will release all code and data when this paper is published.

变压器设计用于处理无资源限制的可变长度输入。然而，当输入超过略大于预训练上下文窗口的阈值时，它们的性能会显著下降。这种对有效上下文窗口的限制限制了基于transformer的大型语言模型（llm）的应用，而这些模型一直是备受期待的主题。因此，预训练的llm的泛化处理不同的输入长度成为一个关键和艰巨的挑战。先前的研究试图通过修改旋转位置嵌入（RoPE）来解决这一挑战，这是处理不同输入长度差异的主要因素。这些努力提供了有价值的见解，但它们通常缺乏对性能下降的根本原因的深刻理解，并且严重依赖手动参数调优。针对这些问题，我们进行了全面的分析，并确定了性能下降背后的两个主要原因：全局分布不匹配和局部分辨率下降。鉴于这些挑战，我们引入了使用贝叶斯优化（BO）的最优RoPE （ORoPE）扩展，这减轻了对额外模型训练的需求。我们的实验证明了我们的方法的有效性，在评估长度为8K、16K和32K时，分别比基线高出21.9%、32.1%和41.2%。本文发表后，我们将发布所有代码和数据。

{"title":"Optimal RoPE extension via Bayesian Optimization for training-free length generalization","authors":"Xinrong Zhang , Shengding Hu , Weilin Zhao , Huadong Wang , Xu Han , Chaoqun He , Guoyang Zeng , Zhiyuan Liu , Maosong Sun","doi":"10.1016/j.aiopen.2025.01.002","DOIUrl":"10.1016/j.aiopen.2025.01.002","url":null,"abstract":"<div><div>Transformers are designed to process input of variable length without resource constraints. However, their performance significantly deteriorates when the input surpasses a threshold slightly larger than the pre-training context window. This limitation on the effective context window confines the application of Transformer-based large language models (LLMs) that have been the subject of great anticipation. Consequently, the generalization of pre-trained LLMs to handle varying input lengths becomes a pivotal and formidable challenge. Previous research has endeavored to address this challenge by modifying the Rotary Position Embedding (RoPE), the primary factor responsible for disparities in handling different input lengths. These efforts have provided valuable insights, while they often lack a deep understanding of the root causes of performance degradation and rely heavily on manual parameter tuning. In response to these issues, we conduct a comprehensive analysis and identify two primary causes behind the performance drop: global distribution mismatch and local resolution degradation. In light of these challenges, we introduce an Optimal RoPE (ORoPE) extension using Bayesian Optimization (BO), which alleviates the need for additional model training. Our experiments demonstrate the efficacy of our approach, outperforming baselines by up to 21.9%, 32.1%, and 41.2% at evaluation lengths of 8K, 16K, and 32K, respectively. We will release all code and data when this paper is published.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"6 ","pages":"Pages 1-11"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143134370","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Symbolic learning enables self-evolving agents 符号学习使代理人能够自我进化

IF 14.8

AI Open

Pub Date : 2025-01-01 DOI: 10.1016/j.aiopen.2025.11.004

Yixin Ou , Wangchunshu Zhou , Shengwei Ding , Long Li , Jialong Wu , Tiannan Wang , Jiamin Chen , Shuai Wang , Xiaohua Xu , Ningyu Zhang , Huajun Chen , Yuchen Eleanor Jiang

The AI community has been exploring a pathway to artificial general intelligence (AGI) by developing “language agents”, which are complex large language models (LLMs) workflows involving both prompting techniques and tool usage methods. While language agents have demonstrated impressive capabilities for many real-world tasks, a fundamental limitation of current language agents research is that they are model-centric or engineering-centric. That is to say, the design of prompts, tools, and workflows of language agents requires substantial manual engineering efforts from human experts rather than automatically learning from data. We believe the transition from model-centric, or engineering-centric, to data-centric, i.e., the ability of language agents to autonomously learn and evolve in environments, is the key for them to possibly achieve AGI.

In this work, we introduce agent symbolic learning, a systematic framework that enables language agents to optimize themselves on their own in a data-centric way using symbolic optimizers. Specifically, we consider agents as symbolic networks in which learnable weights are defined by prompts, tools, and the way they are stacked together. Agent symbolic learning is designed to optimize the symbolic network within language agents in a data-centric way by mimicking two fundamental algorithms in connectionist learning: back-propagation and gradient descent. Instead of dealing with numeric weights, agent symbolic learning works with text-based weights, loss, and gradients. We conduct proof-of-concept experiments on both standard benchmarks and complex real-world tasks and show substantial improvements over static agent frameworks and simple prompt/tool optimization methods. In addition, agent symbolic learning enables language agents to update themselves after being created and deployed in the wild, resulting in “self-evolving agents”. We will open-source the agent symbolic learning framework to facilitate future research on data-centric agent learning.

人工智能社区一直在通过开发“语言代理”探索通往人工通用智能（AGI）的途径，语言代理是复杂的大型语言模型（llm）工作流，涉及提示技术和工具使用方法。虽然语言代理已经在许多现实世界的任务中展示了令人印象深刻的能力，但当前语言代理研究的一个基本限制是它们以模型为中心或以工程为中心。也就是说，语言代理的提示、工具和工作流的设计需要来自人类专家的大量手工工程工作，而不是自动从数据中学习。我们认为，从以模型为中心，或以工程为中心，到以数据为中心的转变，即语言代理在环境中自主学习和进化的能力，是他们可能实现AGI的关键。在这项工作中，我们引入了代理符号学习，这是一个系统框架，使语言代理能够使用符号优化器以数据为中心的方式自行优化自己。具体来说，我们将智能体视为符号网络，其中可学习的权重由提示、工具和它们堆叠在一起的方式定义。智能体符号学习旨在通过模仿连接主义学习中的两种基本算法：反向传播和梯度下降，以数据为中心的方式优化语言智能体中的符号网络。代理符号学习不是处理数字权重，而是处理基于文本的权重、损失和梯度。我们在标准基准测试和复杂的现实世界任务上进行了概念验证实验，并展示了相对于静态代理框架和简单的提示/工具优化方法的实质性改进。此外，智能体符号学习使语言智能体在被创建并部署到野外后能够自我更新，从而形成“自我进化的智能体”。我们将开源代理符号学习框架，以促进未来以数据为中心的代理学习的研究。

{"title":"Symbolic learning enables self-evolving agents","authors":"Yixin Ou , Wangchunshu Zhou , Shengwei Ding , Long Li , Jialong Wu , Tiannan Wang , Jiamin Chen , Shuai Wang , Xiaohua Xu , Ningyu Zhang , Huajun Chen , Yuchen Eleanor Jiang","doi":"10.1016/j.aiopen.2025.11.004","DOIUrl":"10.1016/j.aiopen.2025.11.004","url":null,"abstract":"<div><div>The AI community has been exploring a pathway to artificial general intelligence (AGI) by developing “language agents”, which are complex large language models (LLMs) workflows involving both prompting techniques and tool usage methods. While language agents have demonstrated impressive capabilities for many real-world tasks, a fundamental limitation of current language agents research is that they are model-centric or engineering-centric. That is to say, the design of prompts, tools, and workflows of language agents requires substantial manual engineering efforts from human experts rather than automatically learning from data. We believe the transition from model-centric, or engineering-centric, to data-centric, i.e., the ability of language agents to autonomously learn and evolve in environments, is the key for them to possibly achieve AGI.</div><div>In this work, we introduce <em>agent symbolic learning</em>, a systematic framework that enables language agents to optimize themselves on their own in a data-centric way using <em>symbolic optimizers</em>. Specifically, we consider agents as symbolic networks in which learnable weights are defined by prompts, tools, and the way they are stacked together. Agent symbolic learning is designed to optimize the symbolic network within language agents in a <em>data-centric</em> way by mimicking two fundamental algorithms in connectionist learning: back-propagation and gradient descent. Instead of dealing with numeric weights, agent symbolic learning works with text-based weights, loss, and gradients. We conduct proof-of-concept experiments on both standard benchmarks and complex real-world tasks and show substantial improvements over static agent frameworks and simple prompt/tool optimization methods. In addition, agent symbolic learning enables language agents to update themselves after being created and deployed in the wild, resulting in “self-evolving agents”. We will open-source the agent symbolic learning framework to facilitate future research on <em>data-centric</em> agent learning.</div></div>","PeriodicalId":100068,"journal":{"name":"AI Open","volume":"6 ","pages":"Pages 314-322"},"PeriodicalIF":14.8,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145692936","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

IF 14.8

AI Open

Pub Date : 2025-01-01

引用次数: 0

IF 14.8

AI Open

Pub Date : 2025-01-01

引用次数: 0

IF 14.8

AI Open

Pub Date : 2025-01-01

引用次数: 0

IF 14.8

AI Open

Pub Date : 2025-01-01

引用次数: 0

首页上一页

下一页尾页

类型

全部化学•材料生命科学医学物理工程技术环境•农林材料科学地球科学法学管理学化学环境科学与生态学计算机科学教育学经济学农林科学人文科学生物学数学物理与天体物理心理学综合性期刊其他工业工程理学历史学农学文学信息工程

数据库

全部 ACS Publications Elsevier ieeexplore Springer The Royal Society of Chemistry Wiley

期刊

AI Open

全部 Acc. Chem. Res. ACS Applied Bio Materials ACS Appl. Electron. Mater. ACS Appl. Energy Mater. ACS Appl. Mater. Interfaces ACS Appl. Nano Mater. ACS Appl. Polym. Mater. ACS BIOMATER-SCI ENG ACS Catal. ACS Cent. Sci. ACS Chem. Biol. ACS Chemical Health & Safety ACS Chem. Neurosci. ACS Comb. Sci. ACS Earth Space Chem. ACS Energy Lett. ACS Infect. Dis. ACS Macro Lett. ACS Mater. Lett. ACS Med. Chem. Lett. ACS Nano ACS Omega ACS Photonics ACS Sens. ACS Sustainable Chem. Eng. ACS Synth. Biol. Anal. Chem. BIOCHEMISTRY-US Bioconjugate Chem. BIOMACROMOLECULES Chem. Res. Toxicol. Chem. Rev. Chem. Mater. CRYST GROWTH DES ENERG FUEL Environ. Sci. Technol. Environ. Sci. Technol. Lett. Eur. J. Inorg. Chem. IND ENG CHEM RES Inorg. Chem. J. Agric. Food. Chem. J. Chem. Eng. Data J. Chem. Educ. J. Chem. Inf. Model. J. Chem. Theory Comput. J. Med. Chem. J. Nat. Prod. J PROTEOME RES J. Am. Chem. Soc. LANGMUIR MACROMOLECULES Mol. Pharmaceutics Nano Lett. Org. Lett. ORG PROCESS RES DEV ORGANOMETALLICS J. Org. Chem. J. Phys. Chem. J. Phys. Chem. A J. Phys. Chem. B J. Phys. Chem. C J. Phys. Chem. Lett. Analyst Anal. Methods Biomater. Sci. Catal. Sci. Technol. Chem. Commun. Chem. Soc. Rev. CHEM EDUC RES PRACT CRYSTENGCOMM Dalton Trans. Energy Environ. Sci. ENVIRON SCI-NANO ENVIRON SCI-PROC IMP ENVIRON SCI-WAT RES Faraday Discuss. Food Funct. Green Chem. Inorg. Chem. Front. Integr. Biol. J. Anal. At. Spectrom. J. Mater. Chem. A J. Mater. Chem. B J. Mater. Chem. C Lab Chip Mater. Chem. Front. Mater. Horiz. MEDCHEMCOMM Metallomics Mol. Biosyst. Mol. Syst. Des. Eng. Nanoscale Nanoscale Horiz. Nat. Prod. Rep. New J. Chem. Org. Biomol. Chem. Org. Chem. Front. PHOTOCH PHOTOBIO SCI PCCP Polym. Chem.

﹀