Pub Date : 2026-04-01Epub Date: 2026-01-22DOI: 10.1016/j.compeleceng.2026.110942
G.V. Sriramakrishnan , Ashapu Bhavani , V. Srilakshmi , B. Kiran Kumar
Developmental Dysplasia of the Hip (DDH) is a disease which affects newborn babies and young children. In DDH, the acetabulum may be shallow, or the femoral head may not fit correctly, which causes dislocation or instability of the hip joint. The early detection of DDH failed while the symptoms were mild, which led to delayed treatment and caused severe complications. Thus, the Hybrid Pyramid ResNeXt (HP-ResNeXt) is developed for detecting DDH using hip X-radiation (X-ray) images. Hip X-ray images are sourced from a database, and unwanted noise is removed through a Gaussian Adaptive Bilateral Filter (GABF). Then, a noise-free image is passed to the misshapen pelvis landmark detection phase, where Pyramid Non-local UNet (PN-UNet) is used to identify the affected pelvis region. Entropy-based Local Neighborhood Difference Pattern (LNDP) features, and Gray Level Co-Occurrence Matrix (GLCM) are extracted. Finally, the HP-ResNeXt method is applied for DDH detection, which integrates the advantages of Pyramid Network (PyramidNet) and ResNeXt. The newly introduced HP-ResNeXt approach achieved a True Positive Rate (TPR) of 93.272%, a True Negative Rate (TNR) of 92.567%, and an accuracy of 92.588% with a K-value of 8.
{"title":"HP-ResNeXt: Hybrid Pyramid ResNeXt for Detection of Developmental Dysplasia of the Hip in X-ray Image","authors":"G.V. Sriramakrishnan , Ashapu Bhavani , V. Srilakshmi , B. Kiran Kumar","doi":"10.1016/j.compeleceng.2026.110942","DOIUrl":"10.1016/j.compeleceng.2026.110942","url":null,"abstract":"<div><div>Developmental Dysplasia of the Hip (DDH) is a disease which affects newborn babies and young children. In DDH, the acetabulum may be shallow, or the femoral head may not fit correctly, which causes dislocation or instability of the hip joint. The early detection of DDH failed while the symptoms were mild, which led to delayed treatment and caused severe complications. Thus, the Hybrid Pyramid ResNeXt (HP-ResNeXt) is developed for detecting DDH using hip X-radiation (X-ray) images. Hip X-ray images are sourced from a database, and unwanted noise is removed through a Gaussian Adaptive Bilateral Filter (GABF). Then, a noise-free image is passed to the misshapen pelvis landmark detection phase, where Pyramid Non-local UNet (PN-UNet) is used to identify the affected pelvis region. Entropy-based Local Neighborhood Difference Pattern (LNDP) features, and Gray Level Co-Occurrence Matrix (GLCM) are extracted. Finally, the HP-ResNeXt method is applied for DDH detection, which integrates the advantages of Pyramid Network (PyramidNet) and ResNeXt. The newly introduced HP-ResNeXt approach achieved a True Positive Rate (TPR) of 93.272%, a True Negative Rate (TNR) of 92.567%, and an accuracy of 92.588% with a K-value of 8.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110942"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146026041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-24DOI: 10.1016/j.compeleceng.2026.110947
Jihao Zhang, Guangwei Zhang, Ping Li, Chang Liu, Peng Gong
Robust classification of radar signals under low signal-to-noise ratio (SNR) conditions is critical for target recognition, electronic warfare, and radar emitter identification. However, the performance of conventional methods deteriorates severely in noisy environments due to interference and clutter. This paper proposes an effective classification framework based on a Support Vector Machine (SVM) that exploits the joint discriminative power of wavelet entropy and empirical mode decomposition (EMD) entropy. These two entropy measures characterize the intrinsic complexity and time–frequency structure of radar signals corrupted by noise and are combined into a compact two-dimensional feature vector. Extensive experiments on three representative radar modulation types—pulse Doppler (PD), linear frequency modulation (LFM), and pseudo-code phase modulation (PCPM)—demonstrate the robustness of the proposed method over a wide SNR range from −10 dB to 10 dB. The proposed classifier achieves 100% accuracy when the SNR is above 0 dB, maintains 95% accuracy at −5 dB, and still attains 83% accuracy at −10 dB. In comparative tests, it further achieves 56.7% accuracy at −15 dB, outperforming or matching several state-of-the-art SVM-based and deep-learning-based approaches.
{"title":"Classification of radar signals modulation based on SVM using wavelet entropy and empirical mode decomposition entropy","authors":"Jihao Zhang, Guangwei Zhang, Ping Li, Chang Liu, Peng Gong","doi":"10.1016/j.compeleceng.2026.110947","DOIUrl":"10.1016/j.compeleceng.2026.110947","url":null,"abstract":"<div><div>Robust classification of radar signals under low signal-to-noise ratio (SNR) conditions is critical for target recognition, electronic warfare, and radar emitter identification. However, the performance of conventional methods deteriorates severely in noisy environments due to interference and clutter. This paper proposes an effective classification framework based on a Support Vector Machine (SVM) that exploits the joint discriminative power of wavelet entropy and empirical mode decomposition (EMD) entropy. These two entropy measures characterize the intrinsic complexity and time–frequency structure of radar signals corrupted by noise and are combined into a compact two-dimensional feature vector. Extensive experiments on three representative radar modulation types—pulse Doppler (PD), linear frequency modulation (LFM), and pseudo-code phase modulation (PCPM)—demonstrate the robustness of the proposed method over a wide SNR range from −10 dB to 10 dB. The proposed classifier achieves 100% accuracy when the SNR is above 0 dB, maintains 95% accuracy at −5 dB, and still attains 83% accuracy at −10 dB. In comparative tests, it further achieves 56.7% accuracy at −15 dB, outperforming or matching several state-of-the-art SVM-based and deep-learning-based approaches.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110947"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080168","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-30DOI: 10.1016/j.compeleceng.2026.110987
Qian Zheng , Yuyan Han , Yuting Wang , Daqing Liu , Mingxiao Ma , Leilei Meng
This paper investigates the Hybrid Flow Shop Scheduling Problem with Batch Processing Machines and Variable Sublots (HFSP-BVS), considering sequence-dependent setup times and transportation times, with the objective of minimizing total tardiness. The complexity of HFSP-BVS lies in the tight coupling among lot sequencing, lot splitting, and machine assignment, making it highly challenging in modern manufacturing environments. To address this problem, a Mixed-Integer Linear Programming (MILP) model is formulated and validated using the Gurobi solver. Subsequently, a hybrid algorithm, MADDQN_IG, is proposed by integrating the Multi-Agent Double Deep Q-Network (MADDQN) with Iterated Greedy (IG). The algorithm incorporates four key components: (1) a triple two-layer initialization strategy; (2) a dual-layer destruction-reconstruction parameter selection agent; (3) a local search strategy selection agent; and (4) a multi-agent DDQN construction and training process. These elements are embedded within a unified framework to enhance search efficiency and optimization depth. Extensive computational experiments on 100 benchmark instances demonstrate that MADDQN_IG consistently outperforms existing advanced algorithms (NCIG, QABC, vCCEA, GA), achieving superior solution quality and robustness within limited computation time. Specifically, under three termination criteria (δ = 100, 200, 300), MADDQN_IG improves the ARDI by 78.57%–98.57% and ranks first in the Friedman test, confirming the effectiveness and adaptability of the proposed framework.
本文研究了具有批处理机和可变子批的混合流水车间调度问题，考虑了顺序相关的设置时间和运输时间，以最小化总延误为目标。HFSP-BVS的复杂性在于批排序、批拆分和机器分配之间的紧密耦合，这使得它在现代制造环境中极具挑战性。为了解决这个问题，提出了一个混合整数线性规划（MILP）模型，并使用Gurobi求解器进行了验证。随后，将Multi-Agent Double Deep Q-Network （MADDQN）算法与迭代贪婪（IG）算法相结合，提出了一种混合算法MADDQN_IG。该算法包含四个关键部分：(1)三层两层初始化策略；(2)双层破坏重建参数选择剂；(3)局部搜索策略选择代理；(4)多智能体DDQN构建和训练过程。这些元素被嵌入到一个统一的框架中，以提高搜索效率和优化深度。在100个基准实例上的大量计算实验表明，MADDQN_IG持续优于现有的高级算法（NCIG、QABC、vCCEA、GA），在有限的计算时间内实现了卓越的解质量和鲁棒性。具体而言，在三个终止准则（δ = 100,200,300）下，MADDQN_IG将ARDI提高了78.57%-98.57%，在Friedman检验中排名第一，证实了所提框架的有效性和适应性。
{"title":"Optimization of hybrid flow shop scheduling with batch processing and variable sublots via a multi-agent deep reinforcement learning–guided hybrid algorithm","authors":"Qian Zheng , Yuyan Han , Yuting Wang , Daqing Liu , Mingxiao Ma , Leilei Meng","doi":"10.1016/j.compeleceng.2026.110987","DOIUrl":"10.1016/j.compeleceng.2026.110987","url":null,"abstract":"<div><div>This paper investigates the Hybrid Flow Shop Scheduling Problem with Batch Processing Machines and Variable Sublots (HFSP-BVS), considering sequence-dependent setup times and transportation times, with the objective of minimizing total tardiness. The complexity of HFSP-BVS lies in the tight coupling among lot sequencing, lot splitting, and machine assignment, making it highly challenging in modern manufacturing environments. To address this problem, a Mixed-Integer Linear Programming (MILP) model is formulated and validated using the Gurobi solver. Subsequently, a hybrid algorithm, MADDQN_IG, is proposed by integrating the Multi-Agent Double Deep Q-Network (MADDQN) with Iterated Greedy (IG). The algorithm incorporates four key components: (1) a triple two-layer initialization strategy; (2) a dual-layer destruction-reconstruction parameter selection agent; (3) a local search strategy selection agent; and (4) a multi-agent DDQN construction and training process. These elements are embedded within a unified framework to enhance search efficiency and optimization depth. Extensive computational experiments on 100 benchmark instances demonstrate that MADDQN_IG consistently outperforms existing advanced algorithms (NCIG, QABC, vCCEA, GA), achieving superior solution quality and robustness within limited computation time. Specifically, under three termination criteria (<em><strong>δ</strong></em> = 100, 200, 300), MADDQN_IG improves the ARDI by 78.57%–98.57% and ranks first in the Friedman test, confirming the effectiveness and adaptability of the proposed framework.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110987"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080167","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the advancement of technology, cyberattacks on Internet-based services such as email, e-commerce, social networking, and electronic healthcare are increasing. Since many of these services are accessed through URLs, they have become a primary source for cyberattacks, including phishing and malware. Anti-Phishing Working Group (APWG) reported nearly 1 million phishing attacks in the first quarter of 2025. Early detection of malicious URLs is therefore critical to preventing these threats. Therefore, an efficient detection of malicious URLs is an emerging research problem. However, most ML/DL-based studies focus on overall model accuracy and tend to be biased towards majority classes in imbalanced datasets. In this paper, we propose a machine learning-based malicious URL detection framework specifically designed for imbalanced datasets. We use the ISCX-URL2016 dataset to evaluate model performance across multiple ML algorithms and classbalancing techniques. Our proposed framework, combining the LightGBM classifier with ADASYN oversampling, achieves 99.76% accuracy in multi-class and 99.92% in binary classification. Notably, it shows a 5.93% improvement in detecting phishing URLs, a minority class in the dataset, over existing models. A significant achievement of our approach is its uniform performance across all classes, effectively reducing bias towards majority classes, while existing models fail to achieve it, particularly minority classes. We also validated the proposed model using recent datasets. We further evaluate the framework using various feature selection techniques, demonstrating its effectiveness with fewer features. Additionally, we perform statistical significance testing to validate the reliability of our model, confirming its suitability for real-world applications.
{"title":"A framework for handling class imbalance in malicious URL dataset","authors":"K.G. Raghavendra Narayan , Srijanee Mookherji , Vanga Odelu , Rajendra Prasath","doi":"10.1016/j.compeleceng.2026.111004","DOIUrl":"10.1016/j.compeleceng.2026.111004","url":null,"abstract":"<div><div>With the advancement of technology, cyberattacks on Internet-based services such as email, e-commerce, social networking, and electronic healthcare are increasing. Since many of these services are accessed through URLs, they have become a primary source for cyberattacks, including phishing and malware. Anti-Phishing Working Group (APWG) reported nearly 1 million phishing attacks in the first quarter of 2025. Early detection of malicious URLs is therefore critical to preventing these threats. Therefore, an efficient detection of malicious URLs is an emerging research problem. However, most ML/DL-based studies focus on overall model accuracy and tend to be biased towards majority classes in imbalanced datasets. In this paper, we propose a machine learning-based malicious URL detection framework specifically designed for imbalanced datasets. We use the ISCX-URL2016 dataset to evaluate model performance across multiple ML algorithms and classbalancing techniques. Our proposed framework, combining the LightGBM classifier with ADASYN oversampling, achieves 99.76% accuracy in multi-class and 99.92% in binary classification. Notably, it shows a 5.93% improvement in detecting phishing URLs, a minority class in the dataset, over existing models. A significant achievement of our approach is its uniform performance across all classes, effectively reducing bias towards majority classes, while existing models fail to achieve it, particularly minority classes. We also validated the proposed model using recent datasets. We further evaluate the framework using various feature selection techniques, demonstrating its effectiveness with fewer features. Additionally, we perform statistical significance testing to validate the reliability of our model, confirming its suitability for real-world applications.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 111004"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080161","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-28DOI: 10.1016/j.compeleceng.2026.110963
Simon R. Davies, Richard Macfarlane
Ransomware remains one of the most pervasive and disruptive cyber threats, with modern variants employing advanced techniques such as high-speed multithreaded encryption, obfuscation, and intermittent encryption to reduce detection opportunities and accelerate impact. Despite extensive research into detection and mitigation, few studies have systematically quantified the execution performance and behavioural characteristics of contemporary ransomware families. This paper fills this critical gap through a comprehensive, rigorous analysis of 29 active crypto-ransomware strains executed under controlled, isolated conditions.
Two purpose-built datasets were developed: one, a verified ransomware corpus of the most active families, and the other, a structured target dataset emulating enterprise file systems. Controlled executions of each ransomware sample provided robust measurements of total execution time, pre-encryption delay, and encryption performance. Key findings include wide variation in encryption speeds (33 MB/s to 2.79 GB/s), distinct preparatory and encryption sequences, and frequent use of intermittent encryption to maximise throughput and evade detection.
This research presents the first contemporary academic reproducible benchmark of ransomware execution performance. Through the release of these curated datasets and detailed empirical measurements, it addresses a critical gap in understanding ransomware behaviour. The study contributes a publicly accessible ransomware sample dataset, a structured benchmarking dataset, and a comparative performance analysis across major ransomware families. These results reveal how modern ransomware balances speed, stealth, and efficiency, highlighting the rapidly shrinking window for detection and response. The work establishes a rigorous benchmark for advancing research and practical defence development.
{"title":"Comprehensive performance benchmarking and comparative analysis of active ransomware threats","authors":"Simon R. Davies, Richard Macfarlane","doi":"10.1016/j.compeleceng.2026.110963","DOIUrl":"10.1016/j.compeleceng.2026.110963","url":null,"abstract":"<div><div>Ransomware remains one of the most pervasive and disruptive cyber threats, with modern variants employing advanced techniques such as high-speed multithreaded encryption, obfuscation, and intermittent encryption to reduce detection opportunities and accelerate impact. Despite extensive research into detection and mitigation, few studies have systematically quantified the execution performance and behavioural characteristics of contemporary ransomware families. This paper fills this critical gap through a comprehensive, rigorous analysis of 29 active crypto-ransomware strains executed under controlled, isolated conditions.</div><div>Two purpose-built datasets were developed: one, a verified ransomware corpus of the most active families, and the other, a structured target dataset emulating enterprise file systems. Controlled executions of each ransomware sample provided robust measurements of total execution time, pre-encryption delay, and encryption performance. Key findings include wide variation in encryption speeds (33 MB/s to 2.79 GB/s), distinct preparatory and encryption sequences, and frequent use of intermittent encryption to maximise throughput and evade detection.</div><div>This research presents the first contemporary academic reproducible benchmark of ransomware execution performance. Through the release of these curated datasets and detailed empirical measurements, it addresses a critical gap in understanding ransomware behaviour. The study contributes a publicly accessible ransomware sample dataset, a structured benchmarking dataset, and a comparative performance analysis across major ransomware families. These results reveal how modern ransomware balances speed, stealth, and efficiency, highlighting the rapidly shrinking window for detection and response. The work establishes a rigorous benchmark for advancing research and practical defence development.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110963"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080159","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-02-03DOI: 10.1016/j.compeleceng.2026.111006
Shamsul Haq, Prabal Verma
Healthcare 5.0 is a transformative paradigm that revolutionizes healthcare delivery and improves patient outcomes through incorporating cutting-edge technologies. In this alignment, the paper describes the understanding of Healthcare 5.0 involving different emerging technologies and their roles in effective decision outcomes with proper examples. In correspondence to the significance of Healthcare 5.0, the paper is preceded by focusing on the importance of cloud and edge computing in such environments. It also covers different tools and techniques, analytical methods and advanced emerging analytical algorithms for disease management and treatment optimization. Consequently, it examines the applications of edge computing with emerging analytical technologies in healthcare, showcasing various use cases such as remote patient monitoring, personalized medicine, intelligent healthcare systems, and data-driven decision support resulting in improved patient care and operational efficiency. Subsequently, the statistical results with the systematic framework are performed on the basis of 563 papers published in reputed journals and organizations for the comprehensive analysis of existing technologies and to identify research solutions and challenges in the development of Smart Healthcare. Finally, we summarize our key findings and propose future directions for research and smart healthcare development.
{"title":"An extensive examination of adaptive intelligence in cloud-to-edge systems for Healthcare 5.0","authors":"Shamsul Haq, Prabal Verma","doi":"10.1016/j.compeleceng.2026.111006","DOIUrl":"10.1016/j.compeleceng.2026.111006","url":null,"abstract":"<div><div>Healthcare 5.0 is a transformative paradigm that revolutionizes healthcare delivery and improves patient outcomes through incorporating cutting-edge technologies. In this alignment, the paper describes the understanding of Healthcare 5.0 involving different emerging technologies and their roles in effective decision outcomes with proper examples. In correspondence to the significance of Healthcare 5.0, the paper is preceded by focusing on the importance of cloud and edge computing in such environments. It also covers different tools and techniques, analytical methods and advanced emerging analytical algorithms for disease management and treatment optimization. Consequently, it examines the applications of edge computing with emerging analytical technologies in healthcare, showcasing various use cases such as remote patient monitoring, personalized medicine, intelligent healthcare systems, and data-driven decision support resulting in improved patient care and operational efficiency. Subsequently, the statistical results with the systematic framework are performed on the basis of 563 papers published in reputed journals and organizations for the comprehensive analysis of existing technologies and to identify research solutions and challenges in the development of Smart Healthcare. Finally, we summarize our key findings and propose future directions for research and smart healthcare development.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 111006"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174186","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-02-04DOI: 10.1016/j.compeleceng.2026.111022
Monika Srivastava , Gautam Kumar , Rishav Singh
In medical imaging, the segmentation of skin lesions plays a vital role in detecting and treating skin cancer. Deep learning demonstrates its efficacy in this process. However, it largely relies on extensive and well-annotated datasets that are often limited by healthcare agencies privacy restrictions and institutional data silos. Federated Learning (FL) emerged as a boon, enabling collaborative training without sharing data. Yet, in a real-world setting, healthcare bodies may possess various computational capacities that can affect the consistency of the FL framework, posing the requirement of generalising the model architecture. This study proposes a Decentralised Federated Learning (DFL) framework to improve model generalisation for Skin Lesion Segmentation (SkinSeg). It incorporates a novel Voting (V)-based client selection mechanism to identify the most suitable local model based on performance metrics and dataset size. The selected model is then subjected to Gradual Pruning (GP) via a modified Lottery Ticket Hypothesis (LTH) to reduce model complexity while preserving segmentation quality. The pruned model is then broadcast to all clients for further training. The VGPDFL-SkinSeg substantially improved over State-Of-The-Art FL frameworks on benchmark datasets HAM10K, ISIC-2016/17/18 and DermIs+DermQuest. It achieved a client-wise average Dice Coefficient (DSC) of 90.09%, 96.60% Accuracy, 82.45% meanIOU, 13.63% HD95 and 5.20% ASSD. Initially, each client starts with different segmentation models, reflecting practical diverse systems, and gradually converges towards homogeneity. The study shows that gradual pruning up to 40% yields better segmentation quality than fixed pruning at the beginning and is consistent with client scaling.
{"title":"VGPDFL-SkinSeg: Enhancing model generalisation with data diversity via voting-based client selection and gradual pruning for decentralised federated skin lesion segmentation","authors":"Monika Srivastava , Gautam Kumar , Rishav Singh","doi":"10.1016/j.compeleceng.2026.111022","DOIUrl":"10.1016/j.compeleceng.2026.111022","url":null,"abstract":"<div><div>In medical imaging, the segmentation of skin lesions plays a vital role in detecting and treating skin cancer. Deep learning demonstrates its efficacy in this process. However, it largely relies on extensive and well-annotated datasets that are often limited by healthcare agencies privacy restrictions and institutional data silos. Federated Learning (FL) emerged as a boon, enabling collaborative training without sharing data. Yet, in a real-world setting, healthcare bodies may possess various computational capacities that can affect the consistency of the FL framework, posing the requirement of generalising the model architecture. This study proposes a Decentralised Federated Learning (DFL) framework to improve model generalisation for Skin Lesion Segmentation (SkinSeg). It incorporates a novel Voting (V)-based client selection mechanism to identify the most suitable local model based on performance metrics and dataset size. The selected model is then subjected to Gradual Pruning (GP) via a modified Lottery Ticket Hypothesis (LTH) to reduce model complexity while preserving segmentation quality. The pruned model is then broadcast to all clients for further training. The VGPDFL-SkinSeg substantially improved over State-Of-The-Art FL frameworks on benchmark datasets HAM10K, ISIC-2016/17/18 and DermIs+DermQuest. It achieved a client-wise average Dice Coefficient (DSC) of 90.09%, 96.60% Accuracy, 82.45% meanIOU, 13.63% HD95 and 5.20% ASSD. Initially, each client starts with different segmentation models, reflecting practical diverse systems, and gradually converges towards homogeneity. The study shows that gradual pruning up to 40% yields better segmentation quality than fixed pruning at the beginning and is consistent with client scaling.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 111022"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174250","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-20DOI: 10.1016/j.compeleceng.2026.110988
Jingde Jia , Gang Liu , Yifan Li , Rujian Chen , Yisheng Cao , Gang Xiao , Jianchao Tang
The stochastic and intermittent nature of solar energy poses major challenges for photovoltaic (PV) power forecasting. To address this, we propose a Dynamic Weather-Based Forecasting framework (DWBF) that integrates feature principal component analysis (FPCA) with an adaptive encoder–decoder structure. FPCA is employed to reduce dimensionality while preserving key meteorological information. A convolutional neural network (CNN) with a multi-attention mechanism serves as a shared encoder, capturing global dependencies across weather conditions. Based on solar radiation thresholds, input data is classified into sunny, cloudy, and rainy categories, and the model dynamically selects appropriate decoders: a long short-term memory (LSTM) decoder for sunny days to model stable temporal patterns; a transformer decoder for cloudy days to handle nonlinear variations; and a temporal convolutional network (TCN) decoder for rainy days to process sparse and noisy data. Additionally, Gaussian noise smoothing and adaptive interpolation enhance robustness under data-sparse conditions. Experimental results show that the proposed DWBF consistently outperforms traditional single architecture models across multiple metrics, including root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determination (). Overall, DWBF offers a flexible, accurate, and efficient solution for PV power forecasting by combining feature selection, weather-adaptive decoding, and targeted optimization.
{"title":"Photovoltaic power forecasting under dynamic weather conditions: An adaptive encoder–decoder framework with feature dimensionality optimization","authors":"Jingde Jia , Gang Liu , Yifan Li , Rujian Chen , Yisheng Cao , Gang Xiao , Jianchao Tang","doi":"10.1016/j.compeleceng.2026.110988","DOIUrl":"10.1016/j.compeleceng.2026.110988","url":null,"abstract":"<div><div>The stochastic and intermittent nature of solar energy poses major challenges for photovoltaic (PV) power forecasting. To address this, we propose a Dynamic Weather-Based Forecasting framework (DWBF) that integrates feature principal component analysis (FPCA) with an adaptive encoder–decoder structure. FPCA is employed to reduce dimensionality while preserving key meteorological information. A convolutional neural network (CNN) with a multi-attention mechanism serves as a shared encoder, capturing global dependencies across weather conditions. Based on solar radiation thresholds, input data is classified into sunny, cloudy, and rainy categories, and the model dynamically selects appropriate decoders: a long short-term memory (LSTM) decoder for sunny days to model stable temporal patterns; a transformer decoder for cloudy days to handle nonlinear variations; and a temporal convolutional network (TCN) decoder for rainy days to process sparse and noisy data. Additionally, Gaussian noise smoothing and adaptive interpolation enhance robustness under data-sparse conditions. Experimental results show that the proposed DWBF consistently outperforms traditional single architecture models across multiple metrics, including root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and coefficient of determination (<span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span>). Overall, DWBF offers a flexible, accurate, and efficient solution for PV power forecasting by combining feature selection, weather-adaptive decoding, and targeted optimization.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110988"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146001853","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Aging transportation infrastructure worldwide demands innovative artificial intelligence (AI) solutions for maintenance and monitoring. In this paper, we introduce SUD-ROAD, a new high-resolution dataset and methodology aimed at modernizing road infrastructure management through AI-driven inspection. SUD-ROAD is a specialized subset of the Santiago Urban Dataset, spanning 1635 meters of urban roadway and containing 57 million 3D LiDAR points labeled into seven semantic classes (road pavement, lane lines, other road markings, manhole covers, drains, cracks, and patching). Exploiting the near-planarity of road surfaces, we project the 3D point cloud onto 2D grids, allowing state-of-the-art image-based models to replace more complex 3D networks. A ConvNeXt segmentation model trained on these 2D representations attains a mean Intersection-over-Union of 0.74 and overall accuracy of 0.97, accurately detecting both large-scale assets and fine-grained defects critical for early intervention. We also analyzed the impact of intensity and geometric properties on segmentation effectiveness across different categories. By enabling real-time, AI-driven condition assessment, our approach supports proactive repairs, extends asset life, and reduces life-cycle costs—advancing the broader goal of safer and more sustainable transportation infrastructure. The dataset can be accessed at the following repository: https://github.com/msqiu/SUD-Road.
{"title":"AI-driven road inspection with SUD-ROAD: High-resolution LiDAR benchmark and a novel cross-dimensional semantic segmentation pipeline","authors":"Zhouyan Qiu , Arshia Ghasemlou , Joaquín Martínez-Sánchez , Pedro Arias","doi":"10.1016/j.compeleceng.2026.110993","DOIUrl":"10.1016/j.compeleceng.2026.110993","url":null,"abstract":"<div><div>Aging transportation infrastructure worldwide demands innovative artificial intelligence (AI) solutions for maintenance and monitoring. In this paper, we introduce SUD-ROAD, a new high-resolution dataset and methodology aimed at modernizing road infrastructure management through AI-driven inspection. SUD-ROAD is a specialized subset of the Santiago Urban Dataset, spanning 1635 meters of urban roadway and containing 57 million 3D LiDAR points labeled into seven semantic classes (road pavement, lane lines, other road markings, manhole covers, drains, cracks, and patching). Exploiting the near-planarity of road surfaces, we project the 3D point cloud onto 2D grids, allowing state-of-the-art image-based models to replace more complex 3D networks. A ConvNeXt segmentation model trained on these 2D representations attains a mean Intersection-over-Union of 0.74 and overall accuracy of 0.97, accurately detecting both large-scale assets and fine-grained defects critical for early intervention. We also analyzed the impact of intensity and geometric properties on segmentation effectiveness across different categories. By enabling real-time, AI-driven condition assessment, our approach supports proactive repairs, extends asset life, and reduces life-cycle costs—advancing the broader goal of safer and more sustainable transportation infrastructure. The dataset can be accessed at the following repository: <span><span>https://github.com/msqiu/SUD-Road</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110993"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080162","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-04-01Epub Date: 2026-01-30DOI: 10.1016/j.compeleceng.2026.110994
Mays Abukeshek , Mohammed Al-Mhiqani , Simon Parkinson , Saad Khan , George Bearfield
The rapid digitalisation unfolding in railway systems poses new cybersecurity concerns, thereby requiring solutions that will take the necessary steps to defend against tangible emerging threats. This study aims to systematically review the current cybersecurity research landscape within railway systems. Using a systematic protocol, we comprehensively searched five key online databases: IEEE Xplore, Web of Science, Scopus, ACM, and ScienceDirect. These online databases are recognised for their overall broad coverage and the exhibition of relevance to this study's purpose. Our systematic selection process, facilitated through a predetermined set of inclusion and exclusion criteria, resulted in 114 relevant articles. Among them, 51.8% of the articles reviewed also addressed Control System Security Solutions, while 14% of the articles examined Network Security Solutions, and 12.3% addressed Data Protection and Privacy Solutions. 7% of the articles studied Awareness and Training Solutions, while the remaining 14.9% adopted other approaches. Results identified several significant gaps and challenges relating to railway cybersecurity research, which include issues relating to embracing critical technologies, confirming data privacy, and the need for ongoing education and training of railway workers. Additionally, the study indicated a lack of standardised performance measures and the use of testing datasets, which will impact confidence in measuring the effectiveness of cybersecurity solutions. Ultimately, this research paper advances understanding and contributions to the current railway cybersecurity research landscape, while providing critical recommendations for future research. Efforts towards enhancing international collaboration, adopting emergent technologies such as AI and Blockchain and prioritising education and awareness initiatives are some of the most critical emerging next steps related to cybersecurity and resilience of railway systems.
铁路系统的快速数字化发展带来了新的网络安全问题，因此需要采取必要措施来防御切实的新威胁的解决方案。本研究旨在系统回顾当前铁路系统内的网络安全研究现状。使用系统协议，我们全面检索了五个关键在线数据库：IEEE Xplore， Web of Science, Scopus， ACM和ScienceDirect。这些在线数据库因其全面广泛的覆盖范围和与本研究目的相关的展示而得到认可。我们通过一套预先确定的纳入和排除标准，进行了系统的选择过程，产生了114篇相关文章。其中，51.8%的文章涉及控制系统安全解决方案，14%的文章涉及网络安全解决方案，12.3%的文章涉及数据保护和隐私解决方案，7%的文章研究意识和培训解决方案，其余14.9%采用其他方法。结果确定了与铁路网络安全研究相关的几个重大差距和挑战，其中包括与采用关键技术、确认数据隐私以及对铁路工人进行持续教育和培训的必要性有关的问题。此外，该研究表明，缺乏标准化的性能衡量标准和测试数据集的使用，这将影响衡量网络安全解决方案有效性的信心。最后，本研究论文促进了对当前铁路网络安全研究格局的理解和贡献，同时为未来的研究提供了关键建议。努力加强国际合作，采用人工智能和区块链等新兴技术，优先开展教育和提高意识举措，是与网络安全和铁路系统弹性相关的一些最关键的后续步骤。
{"title":"Cybersecurity in intelligent railway systems: Taxonomy, research trends, challenges, and future directions","authors":"Mays Abukeshek , Mohammed Al-Mhiqani , Simon Parkinson , Saad Khan , George Bearfield","doi":"10.1016/j.compeleceng.2026.110994","DOIUrl":"10.1016/j.compeleceng.2026.110994","url":null,"abstract":"<div><div>The rapid digitalisation unfolding in railway systems poses new cybersecurity concerns, thereby requiring solutions that will take the necessary steps to defend against tangible emerging threats. This study aims to systematically review the current cybersecurity research landscape within railway systems. Using a systematic protocol, we comprehensively searched five key online databases: IEEE Xplore, Web of Science, Scopus, ACM, and ScienceDirect. These online databases are recognised for their overall broad coverage and the exhibition of relevance to this study's purpose. Our systematic selection process, facilitated through a predetermined set of inclusion and exclusion criteria, resulted in 114 relevant articles. Among them, 51.8% of the articles reviewed also addressed Control System Security Solutions, while 14% of the articles examined Network Security Solutions, and 12.3% addressed Data Protection and Privacy Solutions. 7% of the articles studied Awareness and Training Solutions, while the remaining 14.9% adopted other approaches. Results identified several significant gaps and challenges relating to railway cybersecurity research, which include issues relating to embracing critical technologies, confirming data privacy, and the need for ongoing education and training of railway workers. Additionally, the study indicated a lack of standardised performance measures and the use of testing datasets, which will impact confidence in measuring the effectiveness of cybersecurity solutions. Ultimately, this research paper advances understanding and contributions to the current railway cybersecurity research landscape, while providing critical recommendations for future research. Efforts towards enhancing international collaboration, adopting emergent technologies such as AI and Blockchain and prioritising education and awareness initiatives are some of the most critical emerging next steps related to cybersecurity and resilience of railway systems.</div></div>","PeriodicalId":50630,"journal":{"name":"Computers & Electrical Engineering","volume":"132 ","pages":"Article 110994"},"PeriodicalIF":4.9,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080164","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号