Expert Systems with Applications最新文献_第5页

Integration of sustainable diet planning problem under neutrosophic fuzzy multi-objective optimization 中性模糊多目标优化下可持续饮食规划问题的集成

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129360

Kumari Divya, Prabjot Kaur

{"title":"Integration of sustainable diet planning problem under neutrosophic fuzzy multi-objective optimization","authors":"Kumari Divya, Prabjot Kaur","doi":"10.1016/j.eswa.2025.129360","DOIUrl":"10.1016/j.eswa.2025.129360","url":null,"abstract":"<div><div>Sustainable diet planning for diabetic individuals requires a complex trade-off among nutritional adequacy, cultural preferences, affordability, and environmental impact under uncertainty. Existing optimization approaches, including intuitionistic fuzzy sets (IFS), capture acceptance and rejection but fail to model indeterminacy, which is an important aspect of real-world dietary decisions. This study proposes a Neutrosophic Fuzzy Multi-objective Optimization (NFO) framework that incorporates truth, indeterminacy, and falsity membership functions to handle hesitancy in nutritional planning. The model optimizes the six conflicting objectives, maximizing the intake of fiber, protein, and carbohydrates and minimizing fat, sugar, and cost, subject to constraints aligned with Indian Council of Medical Research (ICMR) guidelines. A case study involving 11 common Indian food items across different age and gender groups validates the proposed framework. Comparative analysis with an IFS-based model and two variants of the NFO model reveals that NFO Model II consistently yields more balanced and robust diet plans across demographic groups. The proposed approach offers a computationally efficient and adaptable model for personalized diabetic meal planning, with broader implications for public health nutrition and sustainable food policy.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"297 ","pages":"Article 129360"},"PeriodicalIF":7.5,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144996632","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129510

Yi Zhuang , Jiayu Zhang , Yujia Ge , Nan Jiang

{"title":"Fine-grained similarity retrieval of lesion areas in lung CT images based on visual similarity matching of image blocks","authors":"Yi Zhuang , Jiayu Zhang , Yujia Ge , Nan Jiang","doi":"10.1016/j.eswa.2025.129510","DOIUrl":"10.1016/j.eswa.2025.129510","url":null,"abstract":"<div><h3>Background and Objective</h3><div>With the rapid growth in the number of medical images, the need for content- based medical image retrieval (CBMIR) in clinical aid diagnosis is becoming increasingly important. Most current content-based CT image similarity retrieval methods use the entire CT image, ignoring the fact that the localized lesion region is the main target of similarity retrieval;</div></div><div><h3>Methods</h3><div>To address this issue, the paper proposes a fine-grained similarity retrieval method for lung CT images based on image block(<em>IB</em>) similarity matching, taking lung CT images as an example. In this method, two enabling techniques are introduced: 1) a hybrid Convolution and Vision Transformer Model(CVTM) that effectively captures both local texture and global context features of lesion regions; 2) the iDS high-dimensional index designed to accelerate retrieval among <em>IB</em>;</div></div><div><h3>Results</h3><div>With the aid of these techniques, fine-grained similarity retrieval optimization of lung CT images can be achieved, which facilitates more accurate lesion-level comparison and supports clinical decision-making;</div></div><div><h3>Conclusions</h3><div>Extensive experiments are conducted to indicate that the proposed fine-grained similarity retrieval method achieves excellent performance, with a mAP of 91.33%. Meanwhile, the retrieval efficiency of the iDS high-dimensional index is about 150% higher than that of sequential retrieval, especially when the retrieval radius is large and the database size is substantial.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"298 ","pages":"Article 129510"},"PeriodicalIF":7.5,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145021130","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Learning-based trajectory planning for AGVs in dynamic environment 动态环境下基于学习的agv轨迹规划

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129616

Runda Zhang, Zhida Xing, Senchun Chai, Yuanqing Xia, Runqi Chai

引用次数: 0

Contrastive learning and physics oriented evaluation for advanced segmentation in electron tomography 对比学习和面向物理的电子断层成像高级分割评价

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129606

Cyril Li , Christophe Ducottet , Maxime Moreaud , Sylvain Desroziers , Valentina Girelli Consolaro , Virgile Rouchon , Ovidiu Ersen

{"title":"Contrastive learning and physics oriented evaluation for advanced segmentation in electron tomography","authors":"Cyril Li , Christophe Ducottet , Maxime Moreaud , Sylvain Desroziers , Valentina Girelli Consolaro , Virgile Rouchon , Ovidiu Ersen","doi":"10.1016/j.eswa.2025.129606","DOIUrl":"10.1016/j.eswa.2025.129606","url":null,"abstract":"<div><div>Deep learning methods are now achieving strong results for segmentation tasks, and the standard metric for evaluating methods is the Intersection over Union (IOU). However, we show in this paper that IOU is not efficient in evaluating the quality of segmentation for electron tomography (ET) images of zeolites. We perform a physics-oriented evaluation to ensure that the segmentation results yield coherent physical measures. We also formalize Mixed Supervised / Self-Supervised Contrastive Learning Segmentation (M3S-CLS), a semi-supervised approach using a contrastive learning approach that uses expert annotations to train the neural network model. A detailed comparison of this method with a standard cross-entropy-based model is provided. In addition, we publish a database of five fully segmented ET volumes along with corresponding baseline results. The code and the database is available at <span><span>http://gitlab.univ-st-etienne.fr/labhc-iscv/M3S-CLS</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"298 ","pages":"Article 129606"},"PeriodicalIF":7.5,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145057127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Incorporating estimated depth maps and multi-modal pretraining to improve salient object detection in optical remote sensing images 结合估计深度图和多模态预训练改进光学遥感图像中显著目标的检测

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129624

Yuxiang Fu , Wei Fang

{"title":"Incorporating estimated depth maps and multi-modal pretraining to improve salient object detection in optical remote sensing images","authors":"Yuxiang Fu , Wei Fang","doi":"10.1016/j.eswa.2025.129624","DOIUrl":"10.1016/j.eswa.2025.129624","url":null,"abstract":"<div><div>As a burgeoning theme in optical remote sensing image (ORSI) analysis, salient object detection (SOD) plays a vital role in traffic monitoring, agriculture, disaster management, and other fields. However, the existing ORSI-SOD methods are all single-modal (RGB images primarily), which suffer from performance drop when facing complex scenes (e.g., intricate backgrounds, low contrast scenes, and similar objects). To address this challenge, we introduce estimated depth map to complement RGB image in ORSI-SOD for the first time, which provides 3D geometric cues to improve detection accuracy in complex scenes, thus advancing ORSI-SOD from single-modal to multi-modal. Furthermore, we design a novel pretraining framework: multi-modal reconstructed image pretraining (MMRIP) to pretrain SOD model in multi-modal ORSI-SOD. MMRIP initially utilizes a masked autoencoder (MAE) to restore the masked RGB image; subsequently, it feeds the restored RGB image and clean depth map to the SOD model to generate the saliency map, which can help SOD model more effectively integrate cross modal information and extract better feature. Besides, we present a simple RGB-D SOD model, namely SimSOD, which is pretrained by MMRIP for ORSI-SOD. SimSOD has two major components: DFormer (encoder) and MLP head (decoder). Specifically, we first input RGB image and depth data into the encoder to generate four multi-scale features, then use the decoder to fuse these features and yield the prediction result. Without bells and whistles, our proposed method outperforms the state-of-the-art methods on three public ORSI-SOD datasets. The code can be accessed at: <span><span>https://github.com/Voruarn/MMRIP</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"298 ","pages":"Article 129624"},"PeriodicalIF":7.5,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145021117","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Research on multi-objective optimization of multi-endpoint VRP with time window for the distribution of seasonal products by multi-homing heterogeneous fleets 多归航异构机队季节性产品配送多端点VRP多目标优化研究

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129595

Zhang Yanhu , Yan Lijuan , Kong ShuMei , Miao Decheng

{"title":"Research on multi-objective optimization of multi-endpoint VRP with time window for the distribution of seasonal products by multi-homing heterogeneous fleets","authors":"Zhang Yanhu , Yan Lijuan , Kong ShuMei , Miao Decheng","doi":"10.1016/j.eswa.2025.129595","DOIUrl":"10.1016/j.eswa.2025.129595","url":null,"abstract":"<div><div>This paper addresses a novel VRP variant integrating seasonal demand fluctuations, heterogeneous vehicle sources, and multi-endpoint constraints, focusing on the distribution of seasonal products in a steel parts enterprise. It tackles the complex vehicle routing problem with time windows involving heterogeneous fleets, which encompass different vehicle sources (owned and rented), types (fuel-powered and electric), capacities, ranges, and endpoints. To balance enterprise profitability, greenhouse gas emissions, and environmental quality, we develop a mathematical model centered on optimizing distribution costs, greenhouse gas emissions, and vehicle utilization. Drawing inspiration from ancient competitive activities, we propose a novel Huashan Swords Algorithm (HSSA). Through simulations using real enterprise data, we demonstrate the HSSA’s effectiveness, with comparative experiments against existing advanced algorithms highlighting its superiority. Applying the algorithm to design logistics distribution schemes, we conduct in-depth tests considering different customer groups and fuel station distributions. Analyzing the results from the perspectives of profitability, emissions, and environmental quality, we offer targeted operational suggestions for the enterprise based on its situation, geographical characteristics, and fiscal policies. Moreover, we provide recommendations to local governments on fuel station construction and vehicle subsidy policies, contributing practical solutions to both enterprise operations and regional development.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"298 ","pages":"Article 129595"},"PeriodicalIF":7.5,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145057164","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Collaborative scheduling for heterogeneous robots with simultaneous pickup and delivery tasks in order-picking systems 在取货系统中具有同时取货和交付任务的异构机器人协同调度

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-04 DOI: 10.1016/j.eswa.2025.129457

Xingyu Zhang , Xiuli Wu

{"title":"Collaborative scheduling for heterogeneous robots with simultaneous pickup and delivery tasks in order-picking systems","authors":"Xingyu Zhang , Xiuli Wu","doi":"10.1016/j.eswa.2025.129457","DOIUrl":"10.1016/j.eswa.2025.129457","url":null,"abstract":"<div><div>The rapid growth of the e-commerce has intensified demand on the intelligent warehouse efficiency, where the heterogeneous robots collaboration presents both opportunities and challenges. This study addresses the collaborative scheduling for heterogeneous robots with simultaneous pickup and delivery (CSHR-SPD) problem in autonomous mobile robots (AMRs) and autonomous case-handling robots (ACRs) systems. The CSHR-SPD problem involves three-stage sequential operations (ACR-AMR-ACR) with cross-stage resource competition and spatiotemporal constraints. To solve this NP-hard problem, a mathematical model is formulated to minimize makespan, and an improved variable neighborhood search (IVNS) algorithm is proposed. The algorithm incorporates several key innovations. A terminal stage driven initialization method is proposed to alleviate the delayed startup of AMR. Inspired by the taxi reservation mechanism, an idle robot reservation decoding method is introduced to reduce the waiting time of robots. A bottleneck-aware hybrid neighborhood set is proposed: a tail task reassignment operator optimizes the tail tasks that directly restrict the makespan, while a maximum gap operator focuses on eliminating major unproductive time intervals. Numerical experiments demonstrate that the proposed IVNS algorithm solves the CSHR-SPD problem effectively and efficiently.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"297 ","pages":"Article 129457"},"PeriodicalIF":7.5,"publicationDate":"2025-09-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144996631","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

AI-driven 5G-IoT optimization: Q-learning for real-time energy and network resource management ai驱动的5G-IoT优化：面向实时能源和网络资源管理的q学习

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-03 DOI: 10.1016/j.eswa.2025.129619

Bavethra Murthy, Palani Uthirapathy

{"title":"AI-driven 5G-IoT optimization: Q-learning for real-time energy and network resource management","authors":"Bavethra Murthy, Palani Uthirapathy","doi":"10.1016/j.eswa.2025.129619","DOIUrl":"10.1016/j.eswa.2025.129619","url":null,"abstract":"<div><div>In recent times, the integration of the 5G-enabled Internet of Things has revolutionized through high-speed data transmission, ultra-low latency, and interconnectivity of massive devices. However, the proliferation of 5G-enabled Internet of Things introduces major challenges, such as energy inefficiency and unreliable data delivery in the resource constrained Internet of Things devices. This research proposes a novel Q-Learning-based optimization framework tailored to address these challenges by integrating Radio Frequency energy harvesting, adaptive beamforming, and dynamic resource allocation within the massive Multiple-Input-Multiple-Output system. The proposed model utilizes reinforcement learning to manage the network resources including modulation schemes, beamforming, and energy allocation. By modeling the optimization problem as a Markov Decision Process, the proposed framework dynamically adapts to real-time network conditions to enhance energy efficiency, reliable data delivery, and throughput. The experimental validation demonstrates that the Q-Learning-based strategy effectively optimizes the energy efficiency as well as data transmission and achieves a higher energy efficiency of 98.87 %, higher packet delivery ratio of 98.85 %, lower latency of 1.5 ms, and higher throughput of 200Mbps compared to existing methodologies. This result indicates that the proposed Q-Learning-based framework has the potential to enhance the sustainability and reliability of the 5G-enabled Internet of Things.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"298 ","pages":"Article 129619"},"PeriodicalIF":7.5,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145021132","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Image captioning in low resource assamese language with semantic information prior and spatially encoded transformer model 基于语义信息先验和空间编码转换模型的低资源阿萨姆语图像字幕

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-03 DOI: 10.1016/j.eswa.2025.129479

Pankaj Choudhury , Sidharth Nair , Prithwijit Guha , Sukumar Nandi

{"title":"Image captioning in low resource assamese language with semantic information prior and spatially encoded transformer model","authors":"Pankaj Choudhury , Sidharth Nair , Prithwijit Guha , Sukumar Nandi","doi":"10.1016/j.eswa.2025.129479","DOIUrl":"10.1016/j.eswa.2025.129479","url":null,"abstract":"<div><div>Research on automatic image caption generation in low-resource Indian languages is still in its early stages compared to resource-rich languages like English. In this regard, this work presents a novel approach for generating image captions in the Assamese language. The first contribution of this work is the development of a semantic prior guided transformer model. The semantic prior is a feature vector derived from an initial Assamese caption. The semantic prior introduces Assamese language-specific characteristics to the transformer model. This further helps the model to synthesize a further refined Assamese caption. The second contribution of this paper is the proposal of input visual feature refinement using space-aware positional encoding. This enhances the model’s ability to capture spatial relationships among salient image regions. Additionally, a reinforcement learning based training strategy is employed to enhance the model performance. This proposal is benchmarked on COCO-Assamese and Flickr30k-Assamese datasets against four baseline methods.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"297 ","pages":"Article 129479"},"PeriodicalIF":7.5,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144989211","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Resilience assessment of Chinese semiconductor enterprises based on a novel grey wavelet neural differential dynamics model 基于灰色小波神经微分动力学模型的中国半导体企业弹性评估

IF 7.5 1区计算机科学

Expert Systems with Applications Pub Date : 2025-09-03 DOI: 10.1016/j.eswa.2025.129588

Qi Ding , Xinping Xiao , Xiulin Geng , Lin Luo

{"title":"Resilience assessment of Chinese semiconductor enterprises based on a novel grey wavelet neural differential dynamics model","authors":"Qi Ding , Xinping Xiao , Xiulin Geng , Lin Luo","doi":"10.1016/j.eswa.2025.129588","DOIUrl":"10.1016/j.eswa.2025.129588","url":null,"abstract":"<div><div>In light of the escalating global uncertainties, such as the U.S.-China trade war, an in-depth study of enterprise adaptive strategies and resilience in response to external shocks is of significant theoretical and practical importance for enhancing the survival and development capabilities of enterprises amid global economic turmoil. First, this paper selects nine representative semiconductor enterprises in China as research subjects and evaluates their comprehensive effectiveness on a quarterly basis from 2008 to 2023. The evaluation employs the CRITIC-Grey Relational-TOPSIS method across four dimensions: profitability, adaptability, solvency, and scalability. Second, to address abrupt changes in enterprise effectiveness under uncertain environments, this paper proposes a novel Grey Wavelet Differential Neural Network Model (WNOGM) that integrates Neural Ordinary Differential Equation (NODE), Wavelet Neural Network (WNN), and grey system theory. This model is designed to fit enterprise effectiveness data and assess enterprise resilience. By effectively combining these methodologies, the model enhances its capacity to manage volatile and non-stationary data in dynamic environments, demonstrating greater adaptability and stability. Finally, the empirical analysis reveals that high-capital enterprises demonstrate greater resilience and effectively leverage indigenous substitution policies and technological advancements to recover rapidly. In contrast, mid- and low- capital enterprises exhibit varying degrees of vulnerability, highlighting the necessity to enhance their technological capabilities. These findings offer policymakers a valuable foundation for decision-making to promote sustainable growth amid uncertainty.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"297 ","pages":"Article 129588"},"PeriodicalIF":7.5,"publicationDate":"2025-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144932812","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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