Intelligent Systems with Applications最新文献_第10页

Machine learning alloying design of biodegradable zinc alloy for bone implants using XGBoost and Bayesian optimization 基于XGBoost和贝叶斯优化的生物可降解锌合金骨植入物的机器学习合金化设计

Intelligent Systems with Applications

Pub Date : 2025-06-21 DOI: 10.1016/j.iswa.2025.200549

Mohanad Deif , Hani Attar , Mohammad Aljaidi , Ayoub Alsarhan , Dimah Al-Fraihat , Ahmed Solyman

Developing implants using biodegradable materials eliminates the need for secondary surgery, improves both mechanical and biological properties, and enhances biocompatibility. This study proposes a machine learning approach based on Bayesian optimization (BO) and an eXtreme Gradient Boosting (XGBoost) algorithm to design a biodegradable Zinc (Zn) alloy and forecast percentage of elements in the Zn alloy for bone implants. The dataset employed in this study comprised 1182 samples of Zn alloys obtained from supplementary articles from Google Scholar and the mat web database. For forecasting the mechanical parameters Yield Stress (YS), Ductility, and Ultimate Tensile Strength (UTS), the suggested method got maximum R² values of 0.85, 0.87, and 0.81 demonstrating its exceptional predictive capacity. In addition, the model created a Zn biodegradable alloy with UTS of 363.55 Mpa, YS of 318.93 Mpa, and Ductility of 14 %, which are regarded as good mechanical characteristics meet bone implant criteria. The BO-XGBoost model can expedite the production of the proper alloy for several medical applications, saving time, money, and effort.

使用可生物降解材料开发植入物消除了二次手术的需要，改善了机械和生物性能，并增强了生物相容性。本研究提出了一种基于贝叶斯优化（BO）和极限梯度提升（XGBoost）算法的机器学习方法来设计可生物降解的锌（Zn）合金，并预测用于骨植入物的锌合金中元素的百分比。本研究使用的数据集包括从谷歌Scholar和mat web数据库的补充文章中获得的1182个Zn合金样品。对于屈服应力（YS）、延展性（Ductility）和极限抗拉强度（maximum Tensile Strength， UTS）等力学参数的预测，该方法的R2值分别为0.85、0.87和0.81，具有较好的预测能力。此外，该模型制备的锌可生物降解合金的力学性能良好，UTS为363.55 Mpa， YS为318.93 Mpa，延展性为14%，符合植骨标准。BO-XGBoost模型可以加快几种医疗应用的适当合金的生产，节省时间，金钱和精力。

{"title":"Machine learning alloying design of biodegradable zinc alloy for bone implants using XGBoost and Bayesian optimization","authors":"Mohanad Deif , Hani Attar , Mohammad Aljaidi , Ayoub Alsarhan , Dimah Al-Fraihat , Ahmed Solyman","doi":"10.1016/j.iswa.2025.200549","DOIUrl":"10.1016/j.iswa.2025.200549","url":null,"abstract":"<div><div>Developing implants using biodegradable materials eliminates the need for secondary surgery, improves both mechanical and biological properties, and enhances biocompatibility. This study proposes a machine learning approach based on Bayesian optimization (BO) and an eXtreme Gradient Boosting (XGBoost) algorithm to design a biodegradable Zinc (Zn) alloy and forecast percentage of elements in the Zn alloy for bone implants. The dataset employed in this study comprised 1182 samples of Zn alloys obtained from supplementary articles from Google Scholar and the mat web database. For forecasting the mechanical parameters Yield Stress (YS), Ductility, and Ultimate Tensile Strength (UTS), the suggested method got maximum R<sup>2</sup> values of 0.85, 0.87, and 0.81 demonstrating its exceptional predictive capacity. In addition, the model created a Zn biodegradable alloy with UTS of 363.55 Mpa, YS of 318.93 Mpa, and Ductility of 14 %, which are regarded as good mechanical characteristics meet bone implant criteria. The BO-XGBoost model can expedite the production of the proper alloy for several medical applications, saving time, money, and effort.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"27 ","pages":"Article 200549"},"PeriodicalIF":0.0,"publicationDate":"2025-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144490496","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

Anomaly-based intrusion detection system based on SMOTE-IPF, Whale Optimization Algorithm, and ensemble learning 基于SMOTE-IPF、Whale优化算法和集成学习的基于异常的入侵检测系统

Intelligent Systems with Applications

Pub Date : 2025-06-14 DOI: 10.1016/j.iswa.2025.200543

Tibebu Bekele Shana , Neetu Kumari , Mayank Agarwal , Samrat Mondal , Upaka Rathnayake

Nowadays, cybersecurity is a major worldwide problem. Intrusion detection systems (IDS) help guarantee network security by detecting malicious entries from legitimate entries in network traffic data. IDS has considerable potential for detecting dynamic cyber threats, identifying abnormalities, and identifying malicious conduct within the network. In this paper, we propose Machine Learning (ML) models with an emphasis on the Synthetic Minority Over-sampling Technique (SMOTE) with Iterative Partitioning Filter (IPF) for class imbalance and the Whale Optimization Algorithm (WOA) for feature selection. Class imbalance often results in poorly constructed ML models prioritizing the majority class. In addition, the absence of feature selection can lead to higher computational complexity without impacting performance accuracy. This study uses Bagging, AdaBoost, Extreme Gradient Boosting (XGBoost) and Extra Trees Classifier as classification models. The two widely used datasets to assess the proposed method are NLS-KDD and UNSW-NB15. The K-Fold cross-validation technique trains this model to minimize potential overfitting. These models are evaluated based on performance metrics such as accuracy, precision, recall, and F1-score. The experimental results demonstrate that the Extra Trees Classifier significantly outperforms the baseline models and achieves accuracy values of 99.9% for the NSL-KDD dataset and 97% for the UNSW-NB 15 dataset and outperforms all evaluation measures compared to baseline models for multi-classification of the IDS.

当今，网络安全是一个重大的世界性问题。入侵检测系统（IDS）通过检测网络流量数据中合法条目中的恶意条目，保障网络安全。IDS在检测动态网络威胁、识别异常和识别网络中的恶意行为方面具有相当大的潜力。在本文中，我们提出了机器学习（ML）模型，重点是基于迭代划分过滤器（IPF）的合成少数过采样技术（SMOTE）和鲸鱼优化算法（WOA）的特征选择。类不平衡通常会导致构造不良的ML模型优先考虑大多数类。此外，缺少特征选择可能导致更高的计算复杂度，而不会影响性能准确性。本研究使用Bagging、AdaBoost、Extreme Gradient Boosting （XGBoost）和Extra Trees Classifier作为分类模型。两个广泛使用的数据集是NLS-KDD和UNSW-NB15。K-Fold交叉验证技术训练该模型以最小化潜在的过拟合。这些模型是基于诸如准确性、精度、召回率和f1分数等性能指标进行评估的。实验结果表明，Extra Trees分类器显著优于基线模型，在NSL-KDD数据集和UNSW-NB 15数据集的准确率分别达到99.9%和97%，并且在IDS多分类方面优于基线模型的所有评估指标。

{"title":"Anomaly-based intrusion detection system based on SMOTE-IPF, Whale Optimization Algorithm, and ensemble learning","authors":"Tibebu Bekele Shana , Neetu Kumari , Mayank Agarwal , Samrat Mondal , Upaka Rathnayake","doi":"10.1016/j.iswa.2025.200543","DOIUrl":"10.1016/j.iswa.2025.200543","url":null,"abstract":"<div><div>Nowadays, cybersecurity is a major worldwide problem. Intrusion detection systems (IDS) help guarantee network security by detecting malicious entries from legitimate entries in network traffic data. IDS has considerable potential for detecting dynamic cyber threats, identifying abnormalities, and identifying malicious conduct within the network. In this paper, we propose Machine Learning (ML) models with an emphasis on the Synthetic Minority Over-sampling Technique (SMOTE) with Iterative Partitioning Filter (IPF) for class imbalance and the Whale Optimization Algorithm (WOA) for feature selection. Class imbalance often results in poorly constructed ML models prioritizing the majority class. In addition, the absence of feature selection can lead to higher computational complexity without impacting performance accuracy. This study uses Bagging, AdaBoost, Extreme Gradient Boosting (XGBoost) and Extra Trees Classifier as classification models. The two widely used datasets to assess the proposed method are NLS-KDD and UNSW-NB15. The K-Fold cross-validation technique trains this model to minimize potential overfitting. These models are evaluated based on performance metrics such as accuracy, precision, recall, and F1-score. The experimental results demonstrate that the Extra Trees Classifier significantly outperforms the baseline models and achieves accuracy values of 99.9% for the NSL-KDD dataset and 97% for the UNSW-NB 15 dataset and outperforms all evaluation measures compared to baseline models for multi-classification of the IDS.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"27 ","pages":"Article 200543"},"PeriodicalIF":0.0,"publicationDate":"2025-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144338636","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

Temporal event localization in sports videos via self-supervised proposal generation and cross-modal fusion 基于自监督提案生成和跨模态融合的运动视频时间事件定位

Intelligent Systems with Applications

Pub Date : 2025-06-14 DOI: 10.1016/j.iswa.2025.200539

Guang Xv , Xingchen Wu

Long-form sports videos present unique challenges for temporally localizing relevant segments described by text queries. In this paper, a novel two-stage method is proposed for text snippet localization in long sports videos, combining efficient retrieval with fine-grained refinement. First, an improved video encoding pipeline with a caching mechanism and a video-centric sampling strategy has been designed to efficiently process long videos. Then, a self-supervised proposal generation module is designed that leverages temporal consistency to generate candidate segments (proposals) with pseudo labels, reducing reliance on exhaustive manual annotation. Our model is trained in two stages: a segment-level discrimination stage that learns to identify short video snippets relevant to a query, followed by an instance-level completeness stage that ensures the entire event described by the query is accurately captured. To effectively fuse visual and textual information, a cross-modal fusion strategy is adopted that combines late fusion for scalable coarse retrieval with targeted cross-modal attention for precise alignment. Experiments on sports video datasets demonstrate that our method outperforms state-of-the-art baselines in both accuracy and efficiency.

长篇体育视频对文本查询描述的相关片段的临时本地化提出了独特的挑战。本文提出了一种将高效检索与细粒度细化相结合的两阶段运动视频文本片段定位方法。首先，设计了一种改进的视频编码管道，该管道具有缓存机制和以视频为中心的采样策略，可以有效地处理长视频。然后，设计了一个自监督提案生成模块，该模块利用时间一致性生成具有伪标签的候选片段（提案），减少了对详尽的手动注释的依赖。我们的模型分为两个阶段进行训练：段级识别阶段，学习识别与查询相关的短视频片段，然后是实例级完整性阶段，确保准确捕获查询所描述的整个事件。为了有效地融合视觉和文本信息，采用了一种跨模态融合策略，将后期融合用于可扩展的粗检索与有针对性的跨模态关注相结合，进行精确对齐。在体育视频数据集上的实验表明，我们的方法在准确性和效率方面都优于最先进的基线。

{"title":"Temporal event localization in sports videos via self-supervised proposal generation and cross-modal fusion","authors":"Guang Xv , Xingchen Wu","doi":"10.1016/j.iswa.2025.200539","DOIUrl":"10.1016/j.iswa.2025.200539","url":null,"abstract":"<div><div>Long-form sports videos present unique challenges for temporally localizing relevant segments described by text queries. In this paper, a novel two-stage method is proposed for text snippet localization in long sports videos, combining efficient retrieval with fine-grained refinement. First, an improved video encoding pipeline with a caching mechanism and a video-centric sampling strategy has been designed to efficiently process long videos. Then, a self-supervised proposal generation module is designed that leverages temporal consistency to generate candidate segments (proposals) with pseudo labels, reducing reliance on exhaustive manual annotation. Our model is trained in two stages: a segment-level discrimination stage that learns to identify short video snippets relevant to a query, followed by an instance-level completeness stage that ensures the entire event described by the query is accurately captured. To effectively fuse visual and textual information, a cross-modal fusion strategy is adopted that combines late fusion for scalable coarse retrieval with targeted cross-modal attention for precise alignment. Experiments on sports video datasets demonstrate that our method outperforms state-of-the-art baselines in both accuracy and efficiency.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"27 ","pages":"Article 200539"},"PeriodicalIF":0.0,"publicationDate":"2025-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144514039","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

Situation-aware Cyber–Physical–Social System for Cultural Heritage 情境感知文化遗产的网络-物理-社会系统

Intelligent Systems with Applications

Pub Date : 2025-06-13 DOI: 10.1016/j.iswa.2025.200544

Francesco Colace , Giuseppe D’Aniello , Massimo De Santo , Rosario Gaeta , Gabriel Zuchtriegel

The safeguard of cultural heritage (CH) is one of the most of interest issues for all the countries, like Italy, known for their thousand-year history. Cultural properties have to be maintained regularly and effectively so that the condition of such properties remains good at all times. Human operators have always been the ones in charge of monitoring and maintaining these properties, with domain experts capable of understanding when and how the maintenance has to be done. In our paper, we define a CH asset as a Cyber–Physical–Social System. We designed and proposed a prototype of a Situation-aware Cyber–Physical–Social System (CPSS) for Cultural Heritage, capable of supporting the human operator situation awareness. The CPSS is a Machine Learning (ML) and expert based system equipped with modules for capturing information, which are then processed with ML techniques to identify asset maintenance issues, understanding how they will evolve, and what are the priorities in the maintenance activity to be performed. We propose three case studies relating respectively to: four structures in the archaeological site of Pompeii, three in the archaeological site of Paestum, and three related to the area the archaeological site of the Colosseum, in Rome, for the safeguarding of which the system uses vulnerability indexes, calculated using prior knowledge related to these structures, maintenance issues detected from aerial photos using a YoloV7 detection model, and context space theory with weather and anthropogenic flow data. We showed how it was possible to identify critical and dangerous situations for these zones, with vulnerability indexes capable of mitigating damaged and dangerous areas to be left in that state with the advent of adverse weather phenomena, which indeed from the photos appeared damaged and flooded.

文化遗产的保护是所有国家最关心的问题之一，如意大利，以其千年历史而闻名。必须定期和有效地维护文化财产，使这些财产的状况在任何时候都保持良好。人工操作员一直负责监控和维护这些属性，领域专家能够了解何时以及如何进行维护。在本文中，我们将CH资产定义为一个网络-物理-社会系统。我们设计并提出了一个态势感知的文化遗产网络-物理-社会系统（CPSS）的原型，能够支持人类操作员的态势感知。CPSS是一个基于机器学习（ML）和专家的系统，配备了用于捕获信息的模块，然后用ML技术处理这些信息，以识别资产维护问题，了解它们将如何演变，以及要执行的维护活动中的优先级是什么。我们提出三个案例研究，分别涉及：庞贝考古遗址的4个结构，帕埃斯图姆考古遗址的3个结构，以及罗马斗兽场考古遗址的3个结构，对于这些结构的保护，系统使用脆弱性指数，使用与这些结构相关的先验知识计算，使用YoloV7检测模型从航空照片中检测到的维护问题，以及使用天气和人为流量数据的上下文空间理论。我们展示了如何识别这些区域的关键和危险情况，脆弱性指数能够减轻受损和危险地区在不利天气现象出现时留下的状态，从照片中确实出现了受损和洪水。

{"title":"Situation-aware Cyber–Physical–Social System for Cultural Heritage","authors":"Francesco Colace , Giuseppe D’Aniello , Massimo De Santo , Rosario Gaeta , Gabriel Zuchtriegel","doi":"10.1016/j.iswa.2025.200544","DOIUrl":"10.1016/j.iswa.2025.200544","url":null,"abstract":"<div><div>The safeguard of cultural heritage (CH) is one of the most of interest issues for all the countries, like Italy, known for their thousand-year history. Cultural properties have to be maintained regularly and effectively so that the condition of such properties remains good at all times. Human operators have always been the ones in charge of monitoring and maintaining these properties, with domain experts capable of understanding when and how the maintenance has to be done. In our paper, we define a CH asset as a Cyber–Physical–Social System. We designed and proposed a prototype of a Situation-aware Cyber–Physical–Social System (CPSS) for Cultural Heritage, capable of supporting the human operator situation awareness. The CPSS is a Machine Learning (ML) and expert based system equipped with modules for capturing information, which are then processed with ML techniques to identify asset maintenance issues, understanding how they will evolve, and what are the priorities in the maintenance activity to be performed. We propose three case studies relating respectively to: four structures in the archaeological site of Pompeii, three in the archaeological site of Paestum, and three related to the area the archaeological site of the Colosseum, in Rome, for the safeguarding of which the system uses vulnerability indexes, calculated using prior knowledge related to these structures, maintenance issues detected from aerial photos using a YoloV7 detection model, and context space theory with weather and anthropogenic flow data. We showed how it was possible to identify critical and dangerous situations for these zones, with vulnerability indexes capable of mitigating damaged and dangerous areas to be left in that state with the advent of adverse weather phenomena, which indeed from the photos appeared damaged and flooded.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"27 ","pages":"Article 200544"},"PeriodicalIF":0.0,"publicationDate":"2025-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144298090","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

Advancements and challenges of deep learning architectures for aerial image analysis: A systematic review 航空图像分析中深度学习架构的进步与挑战：系统综述

Intelligent Systems with Applications

Pub Date : 2025-06-04 DOI: 10.1016/j.iswa.2025.200537

Hashibul Ahsan Shoaib , Hadiur Rahman Nabil , Md Anisur Rahman , Md Mohsin Kabir , M.F. Mridha , Jungpil Shin

The rapid advancement of deep learning (DL) technologies has significantly transformed the domain of aerial image analysis. This systematic review explores the forefront of deep learning architectures specifically designed for the processing and analysis of aerial imagery. It offers a comprehensive examination of updated models, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Generative Adversarial Networks (GANs), and Transformers, highlighting their unique contributions and comparative effectiveness in aerial image analysis. This review critically compares these architectures through an extensive literature survey, focusing on their impact on enhancing accuracy, computational efficiency, and overall performance in critical aerial imaging tasks, such as classification, object detection, and semantic segmentation. Additionally, it sheds light on the innovative architectural improvements that have been crucial in overcoming traditional challenges associated with aerial image processing, such as handling high-resolution data, managing diverse and changing landscapes, and ensuring real-time analysis capabilities. By synthesizing current findings and identifying prevailing trends, this review not only charts the progress in the field but also outlines future research directions, emphasizing the need for more adaptable, robust, and efficient deep-learning solutions to meet the growing demands of aerial image analysis.

深度学习（DL）技术的快速发展极大地改变了航空图像分析领域。本系统综述探讨了专门为航空图像处理和分析设计的深度学习架构的前沿。它提供了一个全面的检查更新的模型，如卷积神经网络（cnn），循环神经网络（rnn），生成对抗网络（gan），和变形，突出他们在航空图像分析的独特贡献和比较有效性。这篇综述通过广泛的文献调查对这些架构进行了批判性的比较，重点关注它们对提高准确性、计算效率和关键航空成像任务（如分类、目标检测和语义分割）的整体性能的影响。此外，它还揭示了创新的建筑改进，这些改进对于克服与航空图像处理相关的传统挑战至关重要，例如处理高分辨率数据，管理多样化和不断变化的景观，以及确保实时分析能力。通过综合目前的研究结果和确定流行趋势，本文不仅概述了该领域的进展，还概述了未来的研究方向，强调需要更具适应性、鲁棒性和效率的深度学习解决方案，以满足日益增长的航空图像分析需求。

{"title":"Advancements and challenges of deep learning architectures for aerial image analysis: A systematic review","authors":"Hashibul Ahsan Shoaib , Hadiur Rahman Nabil , Md Anisur Rahman , Md Mohsin Kabir , M.F. Mridha , Jungpil Shin","doi":"10.1016/j.iswa.2025.200537","DOIUrl":"10.1016/j.iswa.2025.200537","url":null,"abstract":"<div><div>The rapid advancement of deep learning (DL) technologies has significantly transformed the domain of aerial image analysis. This systematic review explores the forefront of deep learning architectures specifically designed for the processing and analysis of aerial imagery. It offers a comprehensive examination of updated models, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Generative Adversarial Networks (GANs), and Transformers, highlighting their unique contributions and comparative effectiveness in aerial image analysis. This review critically compares these architectures through an extensive literature survey, focusing on their impact on enhancing accuracy, computational efficiency, and overall performance in critical aerial imaging tasks, such as classification, object detection, and semantic segmentation. Additionally, it sheds light on the innovative architectural improvements that have been crucial in overcoming traditional challenges associated with aerial image processing, such as handling high-resolution data, managing diverse and changing landscapes, and ensuring real-time analysis capabilities. By synthesizing current findings and identifying prevailing trends, this review not only charts the progress in the field but also outlines future research directions, emphasizing the need for more adaptable, robust, and efficient deep-learning solutions to meet the growing demands of aerial image analysis.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"27 ","pages":"Article 200537"},"PeriodicalIF":0.0,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144272578","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

A review of neuro-symbolic AI integrating reasoning and learning for advanced cognitive systems 先进认知系统中集成推理和学习的神经符号人工智能研究综述

Intelligent Systems with Applications

Pub Date : 2025-06-01 DOI: 10.1016/j.iswa.2025.200541

Uzma Nawaz , Mufti Anees-ur-Rahaman , Zubair Saeed

Neuro-symbolic AI represents the convergence of two principal paradigms in artificial intelligence: neural networks, which are efficient in data-driven learning, and symbolic reasoning, which offers explainability and logical inference. This hybrid methodology combines the adaptability of neural networks with symbolic AI's interpretability and formal reasoning abilities, which provide a practical framework for advanced cognitive systems. This paper analyzes the present condition of neuro-symbolic AI, emphasizing essential techniques that combine reasoning and learning. We explore models such as Logic Tensor Networks, Differentiable Logic Programs, and Neural Theorem Provers. The study analyzes their impact on the advancement of cognitive systems in natural language processing, robotics, and decision-making. The paper examines the challenges faced by neuro-symbolic AI, such as scalability, integration with multimodal data, and maintaining interpretability without compromising efficiency. By evaluating the strengths and weaknesses of many methodologies, we comprehensively understand the field's development and its potential to revolutionize intelligent systems. In addition, we identify emerging research areas, including the incorporation of ethical frameworks and the development of adaptive dynamic neuro-symbolic systems that respond in real-time. This review aims to guide future research by providing insights into the potential of neuro-symbolic AI to influence the development of the next generation of intelligent, explainable, and adaptive systems.

神经符号人工智能代表了人工智能中两个主要范式的融合：神经网络，它在数据驱动的学习中是高效的，而符号推理，它提供了可解释性和逻辑推理。这种混合方法将神经网络的适应性与符号人工智能的可解释性和形式推理能力相结合，为高级认知系统提供了实用框架。本文分析了神经符号人工智能的现状，强调了推理和学习相结合的基本技术。我们将探讨逻辑张量网络、可微逻辑程序和神经定理证明等模型。该研究分析了它们对自然语言处理、机器人和决策方面的认知系统进步的影响。本文研究了神经符号人工智能面临的挑战，如可扩展性，与多模态数据的集成，以及在不影响效率的情况下保持可解释性。通过评估许多方法的优缺点，我们全面了解该领域的发展及其革命性智能系统的潜力。此外，我们还确定了新兴的研究领域，包括伦理框架的结合和实时响应的自适应动态神经符号系统的发展。这篇综述旨在通过提供对神经符号人工智能影响下一代智能、可解释和自适应系统发展潜力的见解来指导未来的研究。

{"title":"A review of neuro-symbolic AI integrating reasoning and learning for advanced cognitive systems","authors":"Uzma Nawaz , Mufti Anees-ur-Rahaman , Zubair Saeed","doi":"10.1016/j.iswa.2025.200541","DOIUrl":"10.1016/j.iswa.2025.200541","url":null,"abstract":"<div><div>Neuro-symbolic AI represents the convergence of two principal paradigms in artificial intelligence: neural networks, which are efficient in data-driven learning, and symbolic reasoning, which offers explainability and logical inference. This hybrid methodology combines the adaptability of neural networks with symbolic AI's interpretability and formal reasoning abilities, which provide a practical framework for advanced cognitive systems. This paper analyzes the present condition of neuro-symbolic AI, emphasizing essential techniques that combine reasoning and learning. We explore models such as Logic Tensor Networks, Differentiable Logic Programs, and Neural Theorem Provers. The study analyzes their impact on the advancement of cognitive systems in natural language processing, robotics, and decision-making. The paper examines the challenges faced by neuro-symbolic AI, such as scalability, integration with multimodal data, and maintaining interpretability without compromising efficiency. By evaluating the strengths and weaknesses of many methodologies, we comprehensively understand the field's development and its potential to revolutionize intelligent systems. In addition, we identify emerging research areas, including the incorporation of ethical frameworks and the development of adaptive dynamic neuro-symbolic systems that respond in real-time. This review aims to guide future research by providing insights into the potential of neuro-symbolic AI to influence the development of the next generation of intelligent, explainable, and adaptive systems.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200541"},"PeriodicalIF":0.0,"publicationDate":"2025-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144196102","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

Digital twin technology in wind turbine components: A review 风力发电机部件中的数字孪生技术综述

Intelligent Systems with Applications

Pub Date : 2025-06-01 DOI: 10.1016/j.iswa.2025.200535

Jersson X. Leon-Medina , Diego A. Tibaduiza , Núria Parés , Francesc Pozo

The industrial development, the advances in sensor technology and the processing of large amounts of data, have enabled the training and testing of artificial intelligence models that reproduce, with high accuracy, the behavior of some variables of interest. With the consolidation of the big data era and the proliferation of sensors that can acquire information directly from various components of a wind turbine (WT), a digital twin (DT) allows to close the gap between the physical and the digital worlds. It combines historical data, sensor readings, machine learning and physics-based modeling to replicate the behavior of the physical component accurately. This DT can simulate the performance and behavior of the physical object under different conditions and situations, allowing for predicting failures in WT components and determining their remaining useful life. This review describes the existing literature related to the use of DTs and their developments for WT applications and their components in onshore and offshore applications. This review explores various types of DTs and their approaches, aiming to cover different methods of data processing and concepts related to each DT framework. In addition, it identifies insights from various studies and reviews, particularly focusing on the components of WTs.

工业发展、传感器技术的进步和大量数据的处理，使人工智能模型的训练和测试成为可能，这些模型可以高精度地再现一些感兴趣的变量的行为。随着大数据时代的巩固和传感器的扩散，可以直接从风力涡轮机的各个组件（WT）获取信息，数字孪生（DT）可以缩小物理世界和数字世界之间的差距。它结合了历史数据、传感器读数、机器学习和基于物理的建模，以准确地复制物理组件的行为。此DT可以模拟物理对象在不同条件和情况下的性能和行为，从而预测WT组件中的故障并确定其剩余使用寿命。这篇综述描述了现有的与dt的使用相关的文献，以及它们在陆上和海上应用中的WT应用及其组件的发展。这篇综述探讨了各种类型的数据分析及其方法，旨在涵盖与每个数据分析框架相关的不同数据处理方法和概念。此外，它还确定了来自各种研究和评论的见解，特别是关注wt的组成部分。

{"title":"Digital twin technology in wind turbine components: A review","authors":"Jersson X. Leon-Medina , Diego A. Tibaduiza , Núria Parés , Francesc Pozo","doi":"10.1016/j.iswa.2025.200535","DOIUrl":"10.1016/j.iswa.2025.200535","url":null,"abstract":"<div><div>The industrial development, the advances in sensor technology and the processing of large amounts of data, have enabled the training and testing of artificial intelligence models that reproduce, with high accuracy, the behavior of some variables of interest. With the consolidation of the big data era and the proliferation of sensors that can acquire information directly from various components of a wind turbine (WT), a digital twin (DT) allows to close the gap between the physical and the digital worlds. It combines historical data, sensor readings, machine learning and physics-based modeling to replicate the behavior of the physical component accurately. This DT can simulate the performance and behavior of the physical object under different conditions and situations, allowing for predicting failures in WT components and determining their remaining useful life. This review describes the existing literature related to the use of DTs and their developments for WT applications and their components in onshore and offshore applications. This review explores various types of DTs and their approaches, aiming to cover different methods of data processing and concepts related to each DT framework. In addition, it identifies insights from various studies and reviews, particularly focusing on the components of WTs.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200535"},"PeriodicalIF":0.0,"publicationDate":"2025-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144190364","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

Metaheuristics in automated machine learning: Strategies for optimization 自动机器学习中的元启发式：优化策略

Intelligent Systems with Applications

Pub Date : 2025-06-01 DOI: 10.1016/j.iswa.2025.200532

Francesco Zito , El-Ghazali Talbi , Claudia Cavallaro , Vincenzo Cutello , Mario Pavone

The present work explores the application of Automated Machine Learning techniques, particularly on the optimization of Artificial Neural Networks through hyperparameter tuning. Artificial Neural Networks are widely used across various fields, however building and optimizing them presents significant challenges. By employing an effective hyperparameter tuning, shallow neural networks might become competitive with their deeper counterparts, which in turn makes them more suitable for low-power consumption applications. In our work, we highlight the importance of Hyperparameter Optimization in enhancing neural network performance. We examine various metaheuristic algorithms employed and, in particular, their effectiveness in improving model performance across diverse applications. Despite significant advancements in this area, a comprehensive comparison of these algorithms across different deep learning architectures remains lacking. This work aims to fill this gap by systematically evaluating the performance of metaheuristic algorithms in optimizing hyperparameters and discussing advanced techniques such as parallel computing to adapt metaheuristic algorithms for use in hyperparameter optimization with high-dimensional hyperparameter search space.

目前的工作探讨了自动化机器学习技术的应用，特别是通过超参数调谐对人工神经网络进行优化。人工神经网络广泛应用于各个领域，但构建和优化人工神经网络面临着重大挑战。通过采用有效的超参数调整，浅层神经网络可能会与深层神经网络竞争，这反过来又使它们更适合低功耗应用。在我们的工作中，我们强调了超参数优化在提高神经网络性能方面的重要性。我们研究了所采用的各种元启发式算法，特别是它们在提高不同应用程序的模型性能方面的有效性。尽管在这一领域取得了重大进展，但在不同深度学习架构中对这些算法的全面比较仍然缺乏。本工作旨在通过系统地评估元启发式算法在优化超参数方面的性能，并讨论诸如并行计算等先进技术，以使元启发式算法适用于具有高维超参数搜索空间的超参数优化，从而填补这一空白。

{"title":"Metaheuristics in automated machine learning: Strategies for optimization","authors":"Francesco Zito , El-Ghazali Talbi , Claudia Cavallaro , Vincenzo Cutello , Mario Pavone","doi":"10.1016/j.iswa.2025.200532","DOIUrl":"10.1016/j.iswa.2025.200532","url":null,"abstract":"<div><div>The present work explores the application of Automated Machine Learning techniques, particularly on the optimization of Artificial Neural Networks through hyperparameter tuning. Artificial Neural Networks are widely used across various fields, however building and optimizing them presents significant challenges. By employing an effective hyperparameter tuning, shallow neural networks might become competitive with their deeper counterparts, which in turn makes them more suitable for low-power consumption applications. In our work, we highlight the importance of Hyperparameter Optimization in enhancing neural network performance. We examine various metaheuristic algorithms employed and, in particular, their effectiveness in improving model performance across diverse applications. Despite significant advancements in this area, a comprehensive comparison of these algorithms across different deep learning architectures remains lacking. This work aims to fill this gap by systematically evaluating the performance of metaheuristic algorithms in optimizing hyperparameters and discussing advanced techniques such as parallel computing to adapt metaheuristic algorithms for use in hyperparameter optimization with high-dimensional hyperparameter search space.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200532"},"PeriodicalIF":0.0,"publicationDate":"2025-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144185956","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

YOLO-SR: An optimized convolutional architecture for robust ship detection in SAR Imagery YOLO-SR：一种优化的卷积结构，用于SAR图像的鲁棒船舶检测

Intelligent Systems with Applications

Pub Date : 2025-05-23 DOI: 10.1016/j.iswa.2025.200538

Chi Kien Ha, Hoanh Nguyen, Vu Duc Van

Accurate and efficient ship detection in synthetic aperture radar (SAR) imagery remains a challenging task due to speckle noise, scale variations, and the low contrast of small vessels. In this work, we present YOLO-SR, an enhanced version of YOLOv10 tailored for SAR ship detection, introducing four key innovations: Balanced Detail Fusion (BDF), C2f‐MSDR, DySample, and the Focaler-SIoU loss. Our BDF module adaptively merges shallow, fine‐grained features with deeper semantic features, preventing subtle ship signatures from being overshadowed by irrelevant clutter. Concurrently, C2f‐MSDR replaces standard bottleneck layers with multi-scale dilation residual blocks, expanding the receptive field to handle wide variations in ship size. To improve spatial resolution and retain boundary details, we incorporate DySample, a data-driven upsampling strategy that counteracts the artifacts of naive interpolation. Finally, Focaler-SIoU refines bounding-box regression by integrating distance, orientation, shape, and a focal-like reweighting, thereby emphasizing difficult, small, or partially occluded ships. Experimental results on SAR ship detection datasets confirm that YOLO-SR outperforms state-of-the-art methods in both precision and recall, while retaining competitive inference speeds. These advances offer a robust framework for real-time maritime surveillance, enhancing the detection of both small and large ships under challenging SAR conditions.

在合成孔径雷达（SAR）图像中，由于散斑噪声、尺度变化和小船只的低对比度，准确、高效的船舶检测仍然是一项具有挑战性的任务。在这项工作中，我们提出了YOLO-SR，这是为SAR船舶检测量身定制的YOLOv10的增强版本，引入了四个关键创新：平衡细节融合（BDF）， C2f‐MSDR， dyssample和Focaler-SIoU损失。我们的BDF模块自适应地将浅的、细粒度的特征与更深的语义特征合并，防止细微的船舶特征被无关的杂波所掩盖。同时，C2f‐MSDR用多尺度膨胀残余块取代了标准瓶颈层，扩大了接受域，以处理船舶尺寸的广泛变化。为了提高空间分辨率并保留边界细节，我们采用了DySample，这是一种数据驱动的上采样策略，可以抵消原始插值的伪像。最后，focer - siou通过整合距离、方向、形状和类似焦点的重加权来改进边界盒回归，从而强调困难的、小的或部分遮挡的船只。在SAR船舶检测数据集上的实验结果证实，YOLO-SR在精度和召回率方面都优于最先进的方法，同时保持了有竞争力的推理速度。这些进步为实时海上监视提供了一个强大的框架，增强了在具有挑战性的SAR条件下对小型和大型船舶的检测。

{"title":"YOLO-SR: An optimized convolutional architecture for robust ship detection in SAR Imagery","authors":"Chi Kien Ha, Hoanh Nguyen, Vu Duc Van","doi":"10.1016/j.iswa.2025.200538","DOIUrl":"10.1016/j.iswa.2025.200538","url":null,"abstract":"<div><div>Accurate and efficient ship detection in synthetic aperture radar (SAR) imagery remains a challenging task due to speckle noise, scale variations, and the low contrast of small vessels. In this work, we present YOLO-SR, an enhanced version of YOLOv10 tailored for SAR ship detection, introducing four key innovations: Balanced Detail Fusion (BDF), C2f‐MSDR, DySample, and the Focaler-SIoU loss. Our BDF module adaptively merges shallow, fine‐grained features with deeper semantic features, preventing subtle ship signatures from being overshadowed by irrelevant clutter. Concurrently, C2f‐MSDR replaces standard bottleneck layers with multi-scale dilation residual blocks, expanding the receptive field to handle wide variations in ship size. To improve spatial resolution and retain boundary details, we incorporate DySample, a data-driven upsampling strategy that counteracts the artifacts of naive interpolation. Finally, Focaler-SIoU refines bounding-box regression by integrating distance, orientation, shape, and a focal-like reweighting, thereby emphasizing difficult, small, or partially occluded ships. Experimental results on SAR ship detection datasets confirm that YOLO-SR outperforms state-of-the-art methods in both precision and recall, while retaining competitive inference speeds. These advances offer a robust framework for real-time maritime surveillance, enhancing the detection of both small and large ships under challenging SAR conditions.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200538"},"PeriodicalIF":0.0,"publicationDate":"2025-05-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144169914","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

Pedestrian trajectory prediction model based on self-supervised spatiotemporal graph network 基于自监督时空图网络的行人轨迹预测模型

Intelligent Systems with Applications

Pub Date : 2025-05-18 DOI: 10.1016/j.iswa.2025.200533

Shiji Yang, Xuezhong Xiao

To improve the accuracy of pedestrian trajectory prediction, the graph - based pedestrian trajectory modeling method in the pedestrian trajectory prediction scenario is effective. Thus, a pedestrian trajectory prediction model based on a self - supervised spatiotemporal graph network is proposed. Firstly, in the process of spatiotemporal graph modeling, this model introduces hop interaction instead of node interaction to update node features, which greatly reduces the times of graph convolution operations, alleviates the problem of feature smoothing, and greatly improves the accuracy of prediction. Secondly, a unique self-supervised module is added to the model to mine commonalities between pedestrian’s multi-trajectories through self-supervised to further improve the accuracy of prediction. The experiment uses ETH and UCY public datasets to train and evaluate model performance. The experimental results demonstrate that this model exhibits enhancements in both ADE and FDE metrics when compared to the SOTA model, with an average prediction error reduction of 15 % and 10 %, respectively. In scenes with dense pedestrians such as the UNIV dataset, the prediction errors are reduced by 25 % and 22 %.

为了提高行人轨迹预测的准确性，在行人轨迹预测场景中，基于图的行人轨迹建模方法是有效的。为此，提出了一种基于自监督时空图网络的行人轨迹预测模型。首先，该模型在时空图建模过程中，采用跳跃交互代替节点交互来更新节点特征，大大减少了图卷积运算次数，缓解了特征平滑问题，大大提高了预测精度。其次，在模型中加入独特的自监督模块，通过自监督挖掘行人多轨迹之间的共性，进一步提高预测精度；实验使用ETH和UCY公共数据集来训练和评估模型性能。实验结果表明，与SOTA模型相比，该模型在ADE和FDE指标上都有增强，平均预测误差分别降低了15%和10%。在像UNIV数据集这样行人密集的场景中，预测误差分别减少了25%和22%。

{"title":"Pedestrian trajectory prediction model based on self-supervised spatiotemporal graph network","authors":"Shiji Yang, Xuezhong Xiao","doi":"10.1016/j.iswa.2025.200533","DOIUrl":"10.1016/j.iswa.2025.200533","url":null,"abstract":"<div><div>To improve the accuracy of pedestrian trajectory prediction, the graph - based pedestrian trajectory modeling method in the pedestrian trajectory prediction scenario is effective. Thus, a pedestrian trajectory prediction model based on a self - supervised spatiotemporal graph network is proposed. Firstly, in the process of spatiotemporal graph modeling, this model introduces hop interaction instead of node interaction to update node features, which greatly reduces the times of graph convolution operations, alleviates the problem of feature smoothing, and greatly improves the accuracy of prediction. Secondly, a unique self-supervised module is added to the model to mine commonalities between pedestrian’s multi-trajectories through self-supervised to further improve the accuracy of prediction. The experiment uses ETH and UCY public datasets to train and evaluate model performance. The experimental results demonstrate that this model exhibits enhancements in both ADE and FDE metrics when compared to the SOTA model, with an average prediction error reduction of 15 % and 10 %, respectively. In scenes with dense pedestrians such as the UNIV dataset, the prediction errors are reduced by 25 % and 22 %.</div></div>","PeriodicalId":100684,"journal":{"name":"Intelligent Systems with Applications","volume":"26 ","pages":"Article 200533"},"PeriodicalIF":0.0,"publicationDate":"2025-05-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124236","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