Artificial Intelligence最新文献

英文中文

One Mind, Many Tongues: A Deep Dive into Language-Agnostic Knowledge Neurons in Large Language Models 一个头脑，多种语言：深入研究大型语言模型中与语言无关的知识神经元

IF 14.4 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-02-04 DOI: 10.1016/j.artint.2026.104490

Pengfei Cao, Yuheng Chen, Zhuoran Jin, Yubo Chen, Kang Liu, Jun Zhao

引用次数: 0

Predictable artificial intelligence 可预测的人工智能

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-01-30 DOI: 10.1016/j.artint.2026.104491

Lexin Zhou , Pablo A.M. Casares , Fernando Martínez-Plumed , John Burden , Ryan Burnell , Lucy Cheke , Cèsar Ferri , Alexandru Marcoci , Behzad Mehrbakhsh , Yael Moros-Daval , Seán Ó hÉigeartaigh , Danaja Rutar , Wout Schellaert , Konstantinos Voudouris , José Hernández-Orallo

引用次数: 0

Multi-excitation projective simulation with a many-body physics-inspired inductive bias 多体物理感应偏置的多激励投影仿真

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-01-23 DOI: 10.1016/j.artint.2026.104489

Philip A. LeMaitre, Marius Krumm, Hans J. Briegel

With the impressive progress of deep learning, applications relying on machine learning are increasingly being integrated into daily life. However, most deep learning models have an opaque, oracle-like nature that makes it difficult to interpret and understand their decisions. This problem led to the development of the field known as eXplainable Artificial Intelligence (XAI). One method in this field known as Projective Simulation (PS) models a chain-of-thought as a random walk of a particle on a graph with vertices that have concepts attached to them. While this description has various benefits, including the possibility of quantization, it cannot be naturally used to model thoughts that combine several concepts simultaneously. To overcome this limitation, we introduce Multi-Excitation Projective Simulation (MEPS), a generalization that considers a chain-of-thought to be a random walk of several particles on a hypergraph. A definition for a dynamic hypergraph is put forward to describe the agent’s training history along with applications to AI and hypergraph visualization. An inductive bias inspired by the remarkably successful few-body interaction models used in quantum many-body physics is formalized for our classical MEPS framework and employed to tackle the exponential complexity associated with naive implementations of hypergraphs. We prove that our inductive bias reduces the complexity from exponential to polynomial, with the exponent representing the cutoff on the number of particles that can interact. We numerically apply our method to two toy model environments and a more complex scenario that models the diagnosis of a broken computer. These environments demonstrate the resource savings provided by an appropriate choice of the inductive bias, as well as showcasing aspects of interpretability. A quantum model for MEPS is also briefly outlined and some future directions for it are discussed.

随着深度学习令人印象深刻的进步，依赖机器学习的应用越来越多地融入日常生活。然而，大多数深度学习模型具有不透明的、类似于神谕的性质，这使得很难解释和理解它们的决策。这个问题导致了可解释人工智能（XAI）领域的发展。这一领域的一种方法被称为投影模拟（projection Simulation， PS），它将思维链建模为一个粒子在图上的随机游走，图上的顶点带有附加的概念。虽然这种描述有各种好处，包括量化的可能性，但它不能自然地用于同时结合几个概念的思想建模。为了克服这一限制，我们引入了多激励投影模拟（MEPS），这是一种将思想链视为超图上几个粒子的随机游走的推广方法。提出了动态超图的定义来描述智能体的训练历史，以及在人工智能和超图可视化中的应用。由量子多体物理中非常成功的少体相互作用模型所启发的归纳偏置被形式化为我们的经典MEPS框架，并用于解决与超图的朴素实现相关的指数复杂性。我们证明了我们的归纳偏置将复杂性从指数降低到多项式，指数表示可以相互作用的粒子数量的截止。我们在数值上将我们的方法应用于两个玩具模型环境和一个更复杂的场景，即模拟一台损坏的计算机的诊断。这些环境展示了适当选择归纳偏差所节省的资源，并展示了可解释性的各个方面。本文还简要介绍了MEPS的量子模型，并对其未来发展方向进行了讨论。

{"title":"Multi-excitation projective simulation with a many-body physics-inspired inductive bias","authors":"Philip A. LeMaitre, Marius Krumm, Hans J. Briegel","doi":"10.1016/j.artint.2026.104489","DOIUrl":"10.1016/j.artint.2026.104489","url":null,"abstract":"<div><div>With the impressive progress of deep learning, applications relying on machine learning are increasingly being integrated into daily life. However, most deep learning models have an opaque, oracle-like nature that makes it difficult to interpret and understand their decisions. This problem led to the development of the field known as <em>eXplainable Artificial Intelligence</em> (XAI). One method in this field known as <em>Projective Simulation</em> (PS) models a chain-of-thought as a random walk of a particle on a graph with vertices that have concepts attached to them. While this description has various benefits, including the possibility of quantization, it cannot be naturally used to model thoughts that combine several concepts simultaneously. To overcome this limitation, we introduce <em>Multi-Excitation Projective Simulation</em> (MEPS), a generalization that considers a chain-of-thought to be a random walk of several particles on a hypergraph. A definition for a dynamic hypergraph is put forward to describe the agent’s training history along with applications to AI and hypergraph visualization. An inductive bias inspired by the remarkably successful few-body interaction models used in quantum many-body physics is formalized for our classical MEPS framework and employed to tackle the exponential complexity associated with naive implementations of hypergraphs. We prove that our inductive bias reduces the complexity from exponential to polynomial, with the exponent representing the cutoff on the number of particles that can interact. We numerically apply our method to two toy model environments and a more complex scenario that models the diagnosis of a broken computer. These environments demonstrate the resource savings provided by an appropriate choice of the inductive bias, as well as showcasing aspects of interpretability. A quantum model for MEPS is also briefly outlined and some future directions for it are discussed.</div></div>","PeriodicalId":8434,"journal":{"name":"Artificial Intelligence","volume":"352 ","pages":"Article 104489"},"PeriodicalIF":4.6,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033301","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Symbolic pattern planning 符号模式规划

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-01-16 DOI: 10.1016/j.artint.2026.104482

Matteo Cardellini , Enrico Giunchiglia , Marco Maratea

In this paper, we propose a novel approach for solving automated planning problems, called Symbolic Pattern Planning. Given a deterministic planning problem Π, we propose to compute a plan by first fixing a pattern –defined as an arbitrary sequence of actions– and then define a formula encoding the state resulting from the sequential execution of the actions in the pattern, starting from an arbitrary initial state. By allowing each action in the pattern to be executed consecutively zero, one or possibly more times, and by imposing the conditions on the initial and goal states, we can check whether the pattern allows determining a valid plan or whether the pattern needs to be extended and the procedure iterated. We ground our proposal in the numeric planning setting, we prove the correctness and also the completeness of the procedure (provided at each iteration the pattern is extended with a complete sequence of actions), and we define procedures for the pattern selection and for computing quality plans. When exploiting the planning as satisfiability approach, we show that our encoding allows to determine a valid plan in a number of iterations which is never higher than the one needed by the state-of-the-art rolled-up or relaxed-relaxed-∃ symbolic encodings. On the experimental side, we run an extensive analysis which included the problems and systems involved in the numeric track of the 2023 International Planning Competition, showing that the results validate the theoretical findings and that our planner Patty has remarkably good comparative performances.

在本文中，我们提出了一种解决自动化规划问题的新方法，称为符号模式规划。给定一个确定性规划问题Π，我们建议通过首先固定一个模式（定义为一个任意的动作序列）来计算一个计划，然后定义一个公式来编码模式中动作的顺序执行所产生的状态，从一个任意的初始状态开始。通过允许模式中的每个动作连续执行0次、1次或可能多次，并通过对初始状态和目标状态施加条件，我们可以检查模式是否允许确定有效的计划，或者模式是否需要扩展和过程迭代。我们将我们的建议建立在数字规划设置的基础上，我们证明了过程的正确性和完整性（提供在每次迭代中，模式被扩展为一个完整的动作序列），并且我们定义了模式选择和计算质量计划的过程。当利用计划作为可满足性方法时，我们表明，我们的编码允许在多次迭代中确定有效的计划，该计划永远不会高于最先进的卷起或松弛-松弛-∃符号编码所需的计划。在实验方面，我们进行了广泛的分析，其中包括2023年国际规划竞赛数字轨道中涉及的问题和系统，结果验证了理论发现，并且我们的规划师帕蒂具有非常好的比较性能。

{"title":"Symbolic pattern planning","authors":"Matteo Cardellini , Enrico Giunchiglia , Marco Maratea","doi":"10.1016/j.artint.2026.104482","DOIUrl":"10.1016/j.artint.2026.104482","url":null,"abstract":"<div><div>In this paper, we propose a novel approach for solving automated planning problems, called Symbolic Pattern Planning. Given a deterministic planning problem Π, we propose to compute a plan by first fixing a pattern –defined as an arbitrary sequence of actions– and then define a formula encoding the state resulting from the sequential execution of the actions in the pattern, starting from an arbitrary initial state. By allowing each action in the pattern to be executed consecutively zero, one or possibly more times, and by imposing the conditions on the initial and goal states, we can check whether the pattern allows determining a valid plan or whether the pattern needs to be extended and the procedure iterated. We ground our proposal in the numeric planning setting, we prove the correctness and also the completeness of the procedure (provided at each iteration the pattern is extended with a complete sequence of actions), and we define procedures for the pattern selection and for computing quality plans. When exploiting the planning as satisfiability approach, we show that our encoding allows to determine a valid plan in a number of iterations which is never higher than the one needed by the state-of-the-art rolled-up or relaxed-relaxed-∃ symbolic encodings. On the experimental side, we run an extensive analysis which included the problems and systems involved in the numeric track of the 2023 International Planning Competition, showing that the results validate the theoretical findings and that our planner <span>Patty</span> has remarkably good comparative performances.</div></div>","PeriodicalId":8434,"journal":{"name":"Artificial Intelligence","volume":"352 ","pages":"Article 104482"},"PeriodicalIF":4.6,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995523","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-scale signal modulation for variational graph autoencoders 变分图自编码器的多尺度信号调制

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-01-15 DOI: 10.1016/j.artint.2026.104488

Junwei Cheng , Chaobo He , Pengxing Feng , Weixiong Liu , Ke Liang

Variational graph autoencoder (VGAE) has emerged as a prominent framework for graph representation learning, owing to its probabilistic modeling flexibility and expressive power. Most existing VGAE-based methods assume that the posterior distribution of node embeddings follows a Gaussian distribution and rely on graph convolutional networks with low-pass filtering properties to infer the posterior mean and variance. However, our empirical findings reveal that the variance exhibits non-stationary spectral characteristics, indicating a spectral mismatch between the variance signal and the low-pass nature of conventional inference mechanisms. Further analysis reveals a spectral asymmetry between the posterior mean and variance, as they exhibit distinct signal allocations within the node embeddings. Motivated by these insights, we propose MS-VGAE, a novel variational graph autoencoder with multi-scale signal modulation. Specifically, MS-VGAE leverages a Gabor-based wavelet network to perform fine-grained spectral convolution and signal modulation of the posterior variance across multiple frequency scales. Additionally, a compact optimization objective is derived to enable homophily-aware modulation of frequency-band contributions in the variance. To reduce the computational overhead, we develop MS-VGAE_cheby, which replaces the spectral filtering operation in MS-VGAE with a Chebyshev polynomial approximation. Comprehensive experiments results demonstrate that MS-VGAE achieves satisfactory performance on both node clustering and link prediction tasks. Moreover, MS-VGAE_cheby offers an efficient alternative to MS-VGAE, while maintaining comparable performance on large graphs. The code is available at https://github.com/GDM-SCNU/MS-VGAE.

变分图自编码器（VGAE）由于其概率建模的灵活性和表达能力而成为图表示学习的重要框架。现有的基于vga的方法大多假设节点嵌入的后验分布服从高斯分布，并依靠具有低通滤波特性的图卷积网络来推断后验均值和方差。然而，我们的实证研究结果表明，方差表现出非平稳的光谱特征，表明方差信号与传统推理机制的低通性质之间存在光谱失配。进一步的分析揭示了后验均值和方差之间的谱不对称，因为它们在节点嵌入中表现出不同的信号分配。基于这些见解，我们提出了MS-VGAE，一种具有多尺度信号调制的新型变分图自编码器。具体来说，MS-VGAE利用基于gabor的小波网络在多个频率尺度上执行细粒度的频谱卷积和后验方差的信号调制。此外，还推导了一个紧凑的优化目标，以实现方差中频带贡献的同态感知调制。为了减少计算开销，我们开发了MS-VGAEcheby算法，用Chebyshev多项式近似代替MS-VGAE中的频谱滤波操作。综合实验结果表明，MS-VGAE在节点聚类和链路预测任务上都取得了令人满意的性能。此外，MS-VGAEcheby提供了MS-VGAE的有效替代方案，同时在大型图形上保持相当的性能。代码可在https://github.com/GDM-SCNU/MS-VGAE上获得。

{"title":"Multi-scale signal modulation for variational graph autoencoders","authors":"Junwei Cheng , Chaobo He , Pengxing Feng , Weixiong Liu , Ke Liang","doi":"10.1016/j.artint.2026.104488","DOIUrl":"10.1016/j.artint.2026.104488","url":null,"abstract":"<div><div>Variational graph autoencoder (VGAE) has emerged as a prominent framework for graph representation learning, owing to its probabilistic modeling flexibility and expressive power. Most existing VGAE-based methods assume that the posterior distribution of node embeddings follows a Gaussian distribution and rely on graph convolutional networks with low-pass filtering properties to infer the posterior mean and variance. However, our empirical findings reveal that the variance exhibits non-stationary spectral characteristics, indicating a spectral mismatch between the variance signal and the low-pass nature of conventional inference mechanisms. Further analysis reveals a spectral asymmetry between the posterior mean and variance, as they exhibit distinct signal allocations within the node embeddings. Motivated by these insights, we propose MS-VGAE, a novel variational graph autoencoder with multi-scale signal modulation. Specifically, MS-VGAE leverages a Gabor-based wavelet network to perform fine-grained spectral convolution and signal modulation of the posterior variance across multiple frequency scales. Additionally, a compact optimization objective is derived to enable homophily-aware modulation of frequency-band contributions in the variance. To reduce the computational overhead, we develop MS-VGAE<sub><em>cheby</em></sub>, which replaces the spectral filtering operation in MS-VGAE with a Chebyshev polynomial approximation. Comprehensive experiments results demonstrate that MS-VGAE achieves satisfactory performance on both node clustering and link prediction tasks. Moreover, MS-VGAE<sub><em>cheby</em></sub> offers an efficient alternative to MS-VGAE, while maintaining comparable performance on large graphs. The code is available at <span><span>https://github.com/GDM-SCNU/MS-VGAE</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":8434,"journal":{"name":"Artificial Intelligence","volume":"352 ","pages":"Article 104488"},"PeriodicalIF":4.6,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995524","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Settling the score: Portioning with cardinal preferences 解决问题：根据基本偏好进行分配

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-01-15 DOI: 10.1016/j.artint.2026.104487

Edith Elkind , Matthias Greger , Patrick Lederer , Warut Suksompong , Nicholas Teh

We study a portioning setting in which a public resource such as time or money is to be divided among a given set of candidates, and each agent proposes a division of the resource. We consider two families of aggregation rules for this setting—those based on coordinate-wise aggregation and those that optimize some notion of welfare—as well as the recently proposed independent markets rule. We provide a detailed analysis of these rules from an axiomatic perspective, both for classic axioms, such as strategyproofness and Pareto optimality, and for novel axioms, some of which aim to capture proportionality in this setting. Our results indicate that a simple rule that computes the average of the proposals satisfies many of our axioms and fares better than all other considered rules in terms of fairness properties. We complement these results by presenting two characterizations of the average rule.

我们研究了一种分配设置，在这种分配设置中，公共资源（如时间或金钱）被分配给一组给定的候选者，每个agent提出资源的分配。我们考虑了这一设置的两类聚合规则——那些基于坐标智能聚合的规则和那些优化某些福利概念的规则——以及最近提出的独立市场规则。我们从公理的角度对这些规则进行了详细的分析，包括经典公理，如策略证明性和帕累托最优性，以及新公理，其中一些旨在捕捉这种情况下的比例性。我们的结果表明，计算提案平均值的简单规则满足我们的许多公理，并且在公平性方面优于所有其他考虑的规则。我们通过提出平均规则的两个特征来补充这些结果。

引用次数: 0

SHACL validation in the presence of ontologies: Semantics and rewriting techniques 存在本体的SHACL验证：语义和重写技术

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2026-01-13 DOI: 10.1016/j.artint.2026.104483

Anouk Oudshoorn, Magdalena Ortiz, Mantas Šimkus

SHACL and OWL are two prominent W3C standards for managing RDF data. These languages share many features, but they have one fundamental difference: OWL, designed for inferring facts from incomplete data, makes the open-world assumption, whereas SHACL is a constraint language that treats the data as complete and must be validated under the closed-world assumption. The combination of both formalisms is very appealing and has been called for, but their semantic gap is a major challenge, semantically and computationally. In this paper, we advocate a semantics for SHACL validation in the presence of ontologies based on core universal models. We provide a technique for constructing these models for ontologies in the rich data-tractable description logic Horn-

ALCHIQ

. Furthermore, we use a finite representation of this model to develop a rewriting technique that reduces SHACL validation in the presence of ontologies to standard validation. Finally, we study the complexity of SHACL validation in the presence of ontologies, and show that even very simple ontologies make the problem ExpTime-complete, and PTime-complete in data complexity.

acl和OWL是管理RDF数据的两个重要的W3C标准。这两种语言共享许多特性，但它们有一个根本的区别：OWL设计用于从不完整的数据推断事实，采用开放世界假设，而acl是一种约束语言，将数据视为完整，必须在封闭世界假设下进行验证。这两种形式的结合是非常吸引人的，并且已经被呼吁，但是它们的语义差距是一个主要的挑战，在语义和计算上。在本文中，我们提倡在存在基于核心通用模型的本体的情况下进行SHACL验证的语义。我们提供了一种在富数据可处理描述逻辑Horn-ALCHIQ中为本体构建这些模型的技术。此外，我们使用该模型的有限表示来开发一种重写技术，该技术将存在本体时的SHACL验证减少为标准验证。最后，我们研究了存在本体时的acl验证的复杂性，并表明即使是非常简单的本体也会使数据复杂性的ExpTime-complete和PTime-complete问题。

{"title":"SHACL validation in the presence of ontologies: Semantics and rewriting techniques","authors":"Anouk Oudshoorn, Magdalena Ortiz, Mantas Šimkus","doi":"10.1016/j.artint.2026.104483","DOIUrl":"10.1016/j.artint.2026.104483","url":null,"abstract":"<div><div>SHACL and OWL are two prominent W3C standards for managing RDF data. These languages share many features, but they have one fundamental difference: OWL, designed for inferring facts from incomplete data, makes the <em>open-world assumption</em>, whereas SHACL is a constraint language that treats the data as complete and must be validated under the <em>closed-world assumption</em>. The combination of both formalisms is very appealing and has been called for, but their semantic gap is a major challenge, semantically and computationally. In this paper, we advocate a semantics for SHACL validation in the presence of ontologies based on <em>core universal models</em>. We provide a technique for constructing these models for ontologies in the rich data-tractable description logic Horn-<span><math><mi>ALCHIQ</mi></math></span>. Furthermore, we use a finite representation of this model to develop a <em>rewriting technique</em> that reduces SHACL validation in the presence of ontologies to standard validation. Finally, we study the complexity of SHACL validation in the presence of ontologies, and show that even very simple ontologies make the problem <span>ExpTime</span>-complete, and <span>PTime</span>-complete in data complexity.</div></div>","PeriodicalId":8434,"journal":{"name":"Artificial Intelligence","volume":"352 ","pages":"Article 104483"},"PeriodicalIF":4.6,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145962643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Corrigendum to “Kernel-Bounded Clustering: Achieving the Objective of Spectral Clustering without Eigendecomposition” [Artificial Intelligence 350 (2026) 104440] “核有界聚类：实现无特征分解的谱聚类目标”的勘误表[人工智能350 (2026)104440]

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2025-12-31 DOI: 10.1016/j.artint.2025.104473

Hang Zhang , Kai Ming Ting , Ye Zhu

引用次数: 0

Automated planning instance generation with neuro-symbolic AI 使用神经符号AI自动规划实例生成

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2025-12-28 DOI: 10.1016/j.artint.2025.104471

Carlos Núñez-Molina , Pablo Mesejo , Juan Fernández-Olivares

In the field of Automated Planning there is often the need for a set of planning problems from a particular domain, e.g., to be used as training data for Machine Learning methods or as benchmarks in planning competitions. In most cases, these problems are created either by hand or by a domain-specific generator, putting a burden on the human designers. In this paper, we propose NeSIG (Neuro-Symbolic Instance Generator), to the best of our knowledge the first domain-independent method for automatically generating typed-STRIPS planning problems that are valid, diverse and difficult to solve. We formulate problem generation as a Markov Decision Process and train two generative policies with Deep Reinforcement Learning to generate problems with the desired properties. We conduct experiments on five classical domains, comparing our approach against handcrafted, domain-specific instance generators and various ablations. Results show NeSIG is able to automatically generate valid and diverse problems of much greater difficulty (6.8 times more on geometric average) than domain-specific generators, while simultaneously reducing human effort when compared to them. Additionally, it can generalize to problems more than twice the size of those seen during training.

在自动化规划领域，通常需要一组来自特定领域的规划问题，例如，用作机器学习方法的训练数据或作为规划竞赛的基准。在大多数情况下，这些问题要么是手工创建的，要么是由特定于领域的生成器创建的，这给人类设计人员带来了负担。在本文中，我们提出了NeSIG（神经符号实例生成器），据我们所知，这是第一个独立于领域的方法，用于自动生成有效的、多样化的、难以解决的类型条带规划问题。我们将问题生成制定为马尔可夫决策过程，并使用深度强化学习训练两个生成策略来生成具有所需属性的问题。我们在五个经典领域进行了实验，将我们的方法与手工制作的、特定于领域的实例生成器和各种消融进行了比较。结果表明，NeSIG能够自动生成比特定领域生成器难度大得多的有效和多样化的问题（几何平均难度为6.8倍），同时减少了人工工作量。此外，它可以泛化到比训练中看到的问题大一倍以上的问题。

{"title":"Automated planning instance generation with neuro-symbolic AI","authors":"Carlos Núñez-Molina , Pablo Mesejo , Juan Fernández-Olivares","doi":"10.1016/j.artint.2025.104471","DOIUrl":"10.1016/j.artint.2025.104471","url":null,"abstract":"<div><div>In the field of Automated Planning there is often the need for a set of planning problems from a particular domain, e.g., to be used as training data for Machine Learning methods or as benchmarks in planning competitions. In most cases, these problems are created either by hand or by a domain-specific generator, putting a burden on the human designers. In this paper, we propose NeSIG (<em>Neuro-Symbolic Instance Generator</em>), to the best of our knowledge the first domain-independent method for automatically generating typed-STRIPS planning problems that are valid, diverse and difficult to solve. We formulate problem generation as a Markov Decision Process and train two generative policies with Deep Reinforcement Learning to generate problems with the desired properties. We conduct experiments on five classical domains, comparing our approach against handcrafted, domain-specific instance generators and various ablations. Results show NeSIG is able to automatically generate valid and diverse problems of much greater difficulty (6.8 times more on geometric average) than domain-specific generators, while simultaneously reducing human effort when compared to them. Additionally, it can generalize to problems more than twice the size of those seen during training.</div></div>","PeriodicalId":8434,"journal":{"name":"Artificial Intelligence","volume":"352 ","pages":"Article 104471"},"PeriodicalIF":4.6,"publicationDate":"2025-12-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145845507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

LAD2025, A constraint-based solver for the subgraph isomorphism problem 基于约束的子图同构问题求解器LAD2025

IF 4.6 2区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Artificial Intelligence

Pub Date : 2025-12-25 DOI: 10.1016/j.artint.2025.104474

Christine Solnon

The Subgraph Isomorphism Problem (SIP) is an

N P

-complete problem that aims at finding a copy of a pattern graph in a target graph. It may be modelled as a constraint satisfaction problem in a very straightforward way, and exact approaches for solving SIPs usually propagate constraints to reduce the search space. In particular, PathLAD is a solver introduced in 2016 that combines Locally All Different (LAD) constraints with path-based supplemental constraints. In this paper, we introduce LAD2025, which combines a complete refactoring of PathLAD with new features: new supplemental constraints, a weight-based variable ordering heuristic, random restarts with nogood recording, a new value ordering heuristic and a rule for selecting the level of filtering.

子图同构问题（SIP）是一个np完全问题，其目的是在目标图中找到一个模式图的副本。它可以以一种非常直接的方式建模为约束满足问题，并且解决sip的精确方法通常传播约束以减少搜索空间。特别是，PathLAD是2016年推出的一款求解器，它结合了局部所有不同（local All Different， LAD）约束和基于路径的补充约束。在本文中，我们介绍了LAD2025，它结合了PathLAD的完整重构和新的特征：新的补充约束，基于权重的变量排序启发式，无良好记录的随机重启，新的值排序启发式和选择过滤级别的规则。

引用次数: 0

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

Artificial Intelligence

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.

﹀