Artificial Intelligence Review最新文献

{"title":"AI safety landscape for large language models: taxonomy, state-of-the-art, and future directions","authors":"Chen Chen,&nbsp;Xueluan Gong,&nbsp;Ziyao Liu,&nbsp;Weifeng Jiang,&nbsp;Si Qi Goh,&nbsp;Kwok-Yan Lam","doi":"10.1007/s10462-026-11590-x","DOIUrl":"10.1007/s10462-026-11590-x","url":null,"abstract":"<div><p>AI safety is an emerging field of critical importance for the secure adoption and deployment of AI systems. With the recent advancements in large language models (LLMs), the technological landscape surrounding the design, development, and deployment of AI systems has undergone significant change. The failure of AI systems at one organization, or AI risks undertaken by one organization, can propagate down the AI technology supply chain, affect the entire AI ecosystem, and potentially lead to collective failures and cause large-scale harm to society. In this paper, we propose a novel architectural framework for understanding and analyzing AI safety in the context of LLMs, defining its characteristics through three key perspectives: Trustworthy AI, Responsible AI, and Ecosystemic Safe AI. We provide a comprehensive review of current research and advancements in AI safety from these perspectives, identifying major challenges and outlining mitigation strategies. Additionally, we highlight potential future directions that warrant further exploration to advance AI safety research and, ultimately, strengthen public trust in digital transformation.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11590-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837757","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advancing integrated sensing and communications with RL/DRL in 6 G networks: a survey","authors":"Wali Ullah Khan,&nbsp;Waqas Khalid,&nbsp;Muhammad Iqbal,&nbsp;Chiew Foong Kwong,&nbsp;Manzoor Ahmed,&nbsp;Syed Tariq Shah","doi":"10.1007/s10462-026-11610-w","DOIUrl":"10.1007/s10462-026-11610-w","url":null,"abstract":"<div><p>Integrated sensing and communication (ISAC) has emerged as a key enabling technology for sixth-generation (6 G) networks, supporting joint environment perception and data transmission with high spectral efficiency and reduced hardware cost. However, the design and deployment of ISAC systems remain challenging due to dynamic wireless environments, sensing–communication trade-offs, and the increasing complexity of large-scale networks. Reinforcement learning (RL) and deep reinforcement learning (DRL) have recently attracted significant attention as data-driven approaches for addressing these challenges through model-free, adaptive, and real-time optimization. This survey first reviews the fundamentals of ISAC technology and then summarizes major RL and DRL algorithms relevant to ISAC design. It further provides a comprehensive overview of RL/DRL methods for ISAC in 6 G networks. Existing approaches are categorized into value-based, policy-based, and hybrid methods, and are further classified according to representative ISAC scenarios, including reconfigurable intelligent surface (RIS)-assisted systems, unmanned aerial vehicle (UAV) and satellite systems, vehicular networks, and other emerging 6 G applications. The survey highlights reported gains in spectral efficiency, sensing accuracy, adaptability, and robustness, while also identifying key limitations of current approaches. Finally, the survey outlines current challenges, future research directions, and key lessons learned.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11610-w.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Toward a unified governing framework for modern artificial intelligence: a taxonomy-based cross-domain survey","authors":"Abbas Amini,&nbsp;Narjes Firouzkouhi,&nbsp;Sorour Alotaibi,&nbsp;Omar Ali,&nbsp;Maria Rashidi,&nbsp;Ahmad Gholami,&nbsp;Qingbin Zheng,&nbsp;Isam Zabalawi,&nbsp;Chun Cheng,&nbsp;Bijan Davvaz","doi":"10.1007/s10462-026-11596-5","DOIUrl":"10.1007/s10462-026-11596-5","url":null,"abstract":"<div><p>Advanced technologies are experiencing significant transformation through AI (Artificial Intelligence), where mathematical foundations serve as the backbone of classical and modern AI models. Despite rapid advancements and widespread adoption, there is a lack of a unified framework for core AI components, comprising building blocks, governing equations, parameters, evaluation criteria, benchmarks, performance metrics, and objective functions across technological domains. In this comprehensive survey, we address this gap in the areas of energy, renewable energy and water, smart buildings and cities, the environment and climate change, hydrogen and hydrogen fuel cells, and cross-sector advanced technologies, including robotics and autonomous systems, computer vision, finance, and industrial manufacturing, with systematic intercomparison and benchmarking. In addition, we conduct a taxonomy-based analysis with a methodological focus on AI models, their underlying parameters and governing equations, as well as their pros, cons, trade-offs, comparative analyses, and directions for future development. This survey consolidates these elements into a structured reference that defines key requirements for AI development in high-tech sectors and provides a forward-looking roadmap to foster innovation beyond current technological infrastructures.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11596-5.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"From individual decisions to team emergence: a survey on explainable cooperative multi-agent reinforcement learning","authors":"Lei Sheng,&nbsp;Xiliang Chen,&nbsp;Zhiqiang Pan,&nbsp;Fei Cai,&nbsp;Honghui Chen","doi":"10.1007/s10462-026-11598-3","DOIUrl":"10.1007/s10462-026-11598-3","url":null,"abstract":"<div><p>Multi-Agent Reinforcement Learning (MARL) holds significant promise for cooperative decision-making, yet its reliance on deep neural networks (DNNs) creates “black-box” characteristics that impede trustworthy deployment in high-stakes scenarios. This lack of transparency complicates tracing decision logic and raises concerns about safety and accountability. This survey systematically reviews Explainable MARL (XMARL) for cooperative settings, deconstructing the decision-making chain from individual agent policies to collective team behavior. To address the absence of a unified framework, we introduce a novel multi-level taxonomy encompassing microscopic individual behavior, interaction mechanisms, team strategy emergence, and system-level performance. We organize core explanatory questions and technical approaches within this structure, summarize the principles and limitations of representative methods, and critically discuss key challenges such as evaluation standards, causal reasoning integration, and deployment adaptability. Our goal is to provide both theoretical foundation and technical guidance for building transparent, trustworthy, and verifiably cooperative multi-agent systems (MASs).</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11598-3.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Ai for rice leaf disease detection: benchmarks, bias, optimization, and deployment readiness","authors":"Chatter Singh,&nbsp;Amar Singh,&nbsp;Sahraoui Dhelim","doi":"10.1007/s10462-026-11613-7","DOIUrl":"10.1007/s10462-026-11613-7","url":null,"abstract":"<div><p>Rice leaf diseases threaten yield, grain quality, and field-management decisions across intensive and smallholder production systems. Visual scouting remains essential, but it is labor-intensive, subjective, and difficult to scale under variable light, cultivar, growth-stage, and disease-pressure conditions. This article analyzes a curated corpus of 111 image-based rice leaf disease studies identified through a PRISMA-ScR-aligned search conducted during 1 January 2026–31 March 2026, with references restricted to publications available on or before 31 March 2026. The synthesis covers disease symptoms, search methodology, dataset provenance, preprocessing, classical machine learning, deep learning, transformer and hybrid architectures, bio-inspired optimization, explainability, severity estimation, and deployment readiness. Its central contribution is to separate clean benchmark performance from field-relevant evidence by examining data source, split unit, leakage risk, class imbalance, external validation, computational cost, and agronomic usefulness. The analysis shows that high internal accuracy is common on curated three- or four-class datasets, whereas cross-dataset validation, leakage-resistant splitting, class-wise reporting, severity localization, device-level efficiency metrics, and expert-validated explanations remain inconsistent. The article proposes a normalized benchmarking framework, an unweighted study-quality and interpretive-risk rubric, a leakage-aware dataset interpretation scheme, and a standardized reporting checklist for optimization-aware rice disease models. Progress in this area should be judged by reproducible, geographically diverse, severity-aware, explainable, and edge-feasible systems that support practical crop-health decisions, not by marginal gains on clean internal splits alone.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11613-7.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782128","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Coverage and scheduling in directional sensor networks via hybrid mayfly optimization and graph coloring","authors":"Susmitha G,&nbsp;Pavithra R","doi":"10.1007/s10462-026-11601-x","DOIUrl":"10.1007/s10462-026-11601-x","url":null,"abstract":"<p>Directional Sensor Network (DSN) consists of sensors with a limited sensing angle designed to monitor the targets efficiently within a specific field of view. Activating these battery-powered sensors simultaneously depletes their limited energy and affects the overall network performance. Sensor scheduling addresses these limitations by partitioning the working directions of the sensors into disjoint cover sets. Hence, this paper proposes a hybrid Mayfly Optimization and Vertex Coloring Algorithm (MO-VC) by focusing on energy-aware scheduling with optimal sensor deployment. The mayfly optimization algorithm determines the optimal positions of sensors by balancing both exploration and exploitation to maximize the target coverage. The vertex coloring-based scheduling technique constructs directional cover sets which satisfies the coverage requirements while minimizing redundant activities. The efficiency of the proposed algorithm is evaluated using the mathematical upper bound on the maximum number of cover sets. Further, the proposed algorithm is compared with several metaheuristic hybrid algorithms, such as Grey-Wolf Optimization with Vertex Coloring Algorithm, Harris Hawks Optimization with Vertex Coloring Algorithm, Salp Swarm Algorithm with Vertex Coloring, Firefly Algorithm with Vertex Coloring, Arithmetic Optimization Algorithm with Vertex Coloring, and Q-Learning with Vertex Coloring Algorithm. The simulation results show that the proposed MO-VC outperforms in terms of coverage efficiency and determines the maximum number of directional cover sets.</p>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11601-x.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782145","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploring multi-agent learning solutions in modern tabletop games: a survey","authors":"Petra Csereoka,&nbsp;Mihai Victor Micea","doi":"10.1007/s10462-026-11616-4","DOIUrl":"10.1007/s10462-026-11616-4","url":null,"abstract":"<div><p>Games play a key role in human learning by actively engaging people, posing hands-on challenges within a safe and competitive environment. Their interactive components foster skills such as negotiation, strategic thinking, and adaptation to unforeseen events which are key elements for tackling complex challenges. Artificial agents facing non-trivial real-world problems require similar skills, thus tabletop games present a beneficial training environment posing a unique set of challenges that can enhance the performance of artificial agents. This survey provides a systematic and comprehensive review of state-of-the-art training and testing environments implementing tabletop games proposed in recent years, together with the main challenges that were explored, and the Reinforcement Learning approaches implemented to tackle these problems. We begin by outlining how this survey differs from other existing reviews and highlight the main contributions it brings. Then, we analyze and categorize the training environments by different criteria. For each category identified based on the type of interactions that occur between agents, case studies are presented highlighting the modeled problems and algorithms used to enhance the agents’ performance emphasizing the main advantages and disadvantages. Finally, we examine emerging trends, potential applications to other domains, and open challenges, showcasing potential future research directions.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11616-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deepfake detection across image, video, and audio: a comprehensive survey with empirical evaluation of generalization and robustness","authors":"Hong-Hanh Nguyen-Le,&nbsp;Van-Tuan Tran,&nbsp;Dinh-Thuc Nguyen,&nbsp;Nhien-An Le-Khac","doi":"10.1007/s10462-026-11608-4","DOIUrl":"10.1007/s10462-026-11608-4","url":null,"abstract":"<div><p>Deepfakes (DFs) have emerged as a significant threat in recent years. They are exploited for malicious purposes such as impersonation, misinformation dissemination, and artistic style imitation, raising critical ethical and security concerns. This survey presents a comprehensive cross-modality analysis of passive DF detection, examining image, video, and audio modalities simultaneously. Distinct from multimodal detection which fuses multiple data streams (e.g., audio-visual, text-visual), our cross-modality analysis investigates the interconnected relationships, shared methodological principles, and common vulnerabilities across independent modalities. We systematically categorize detection approaches based on their underlying methodologies: forensic-based, data-driven, fingerprint-based, and hybrid techniques for visual modalities, and handcrafted versus learnable features for audio. We also extend our analysis beyond mere detection accuracy to include essential performance dimensions for real-world deployment, including generalization and robustness. Additionally, this survey provides a unified evaluation protocol across 10 popular datasets to assess detection accuracy, generalization, and robustness in each modality. Specifically, we conduct extensive empirical evaluations across three critical dimensions: (1) verification of reported within-domain accuracy for 13 unimodal detectors, (2) cross-domain generalization assessment of 33 methods specifically designed to enhance generalization capability, and (3) robustness evaluation of 6 methods against adversarial attacks. Our experiments reveal a persistent generalization gap, with performance degradations of <span>(10-15%)</span> in out-of-distribution scenarios, and vulnerability to white-box adversarial attacks exceeding <span>(80%)</span> success rates. We also analyze the advantages and limitations of existing datasets, benchmarks, and evaluation metrics for passive DF detection. Finally, we propose future research directions to address these unexplored and emerging issues in the field of passive DF detection. This survey serves as a comprehensive resource for researchers and practitioners, providing insights into the current landscape, methodological approaches, and promising future directions in this rapidly evolving field.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11608-4.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148782490","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable artificial intelligence in neurology: a holistic exploration of models for diagnosis, progression tracking, and treatment planning","authors":"Shiva Toumaj,&nbsp;Arash Heidari","doi":"10.1007/s10462-026-11606-6","DOIUrl":"10.1007/s10462-026-11606-6","url":null,"abstract":"<div><p>Neurological disorders, including neurodegenerative diseases, are among the leading causes of disability and mortality worldwide, with profound effects on cognition, sensory function, motor performance, and overall quality of life. Despite decades of research, effective treatment options remain limited because of complex pathophysiological mechanisms, diagnostic uncertainty, and delayed or insufficiently robust detection. Recent advances in Artificial Intelligence (AI), particularly in Machine Learning (ML) and Deep Learning (DL), have enabled the development of powerful computational models for the diagnosis and management of neurological disorders. However, the opaque nature of many of these black-box models remains a major barrier to clinical adoption, underscoring the need for Explainable Artificial Intelligence (XAI) to enhance trust, transparency, and clinical interpretability. In response to this need, this review presents a comprehensive and task-oriented examination of recent XAI applications and frameworks in neurology. Unlike prior surveys that have focused on isolated neurological conditions, limited model categories, or narrow interpretability issues, this study provides a unified and clinically grounded taxonomy across five major application domains: early diagnosis, disease classification, progression tracking, risk prediction, and treatment planning. The review systematically categorizes existing interpretability techniques and examines their usage trends, evaluation strategies, ethical considerations, clinical relevance, and implementation challenges. It also highlights critical gaps in the current literature, including the absence of explicit security protocols in 96.3% of the reviewed studies. By bridging current research with practical clinical needs, this review aims to provide researchers and clinicians with a clear and structured reference for the development and deployment of explainable models in real-world neurological settings. In doing so, it not only clarifies the current strengths and limitations of XAI in neurology but also identifies important directions for future innovation and responsible clinical integration.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11606-6.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148927677","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deep learning for intravascular optical coherence tomography: a review of plaque analysis","authors":"Yi Chen,&nbsp;Minh-Son To,&nbsp;Harry J. Carpenter,&nbsp;Peter J. Psaltis,&nbsp;Jiawen Li","doi":"10.1007/s10462-026-11607-5","DOIUrl":"10.1007/s10462-026-11607-5","url":null,"abstract":"<div><p>Coronary artery disease, a major global health burden, is characterized by the progressive buildup of plaques in the walls of coronary arteries, leading to restrictive narrowing of blood flow and potentially life-threatening complications. Intravascular optical coherence tomography (OCT) enables high-resolution assessment of plaque features, but image analysis remains time-consuming and operator-dependent. This review examines the application of deep learning (DL) to intravascular OCT for coronary plaque analysis, with the aim of clarifying the current scope of research, synthesizing methodological developments, and identifying the main barriers to clinical translation. We reviewed 80 original studies published between January 2017 and March 2025. The review covers the major methodological components of DL-based plaque analysis, including datasets, preprocessing, validation strategies, evaluation metrics, loss functions, model architectures for classification, detection, and segmentation, as well as post-processing and implementation practices. The literature shows a clear progression from the direct use of general-purpose computer vision models to increasingly specialized architectures and processing pipelines tailored to intravascular OCT data. This evolution also reveals a central tension: although increasing model complexity often improves predictive performance, clinically meaningful deployment still depends on stronger generalizability, robustness, interpretability, and reproducibility. By critically synthesizing these developments, this review provides a structured methodological reference for researchers, highlights the principal technical and translational limitations in the field, and outlines practical directions for developing more reliable and clinically applicable DL systems for intravascular imaging.</p></div>","PeriodicalId":8449,"journal":{"name":"Artificial Intelligence Review","volume":"59 10","pages":""},"PeriodicalIF":18.8,"publicationDate":"2026-06-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10462-026-11607-5.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148837416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}