IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2025-12-02 DOI: 10.1016/j.array.2025.100615

Javed Sayyad, Khush Attarde

Object detection is essential in several industries, including defense, autonomous vehicles, and surveillance. These applications rely on various devices equipped with cameras, such as vehicles, drones, and satellites; primarily operating in the visible spectral domain rather than infrared or other spectral ranges. Deep Learning (DL) techniques have significantly advanced the field of object detection, enabling the identification of various objects. However, detecting tiny objects remains a challenging task. Despite its difficulty, identifying small objects in images captured by these devices in the visible spectrum is crucial. It is essential to explore hybrid techniques and modifications in feature architectures to address the challenge of detecting tiny objects. Simple architectures often fall short in this regard, necessitating more sophisticated approaches. This paper systematically reviews different DL-based approaches researchers have previously employed to tackle this issue. A systematic literature review on SOD and DL techniques uses the ”Preferred Reporting Items for Systematic Reviews and Meta-Analysis” (PRISMA) methodology. It discusses various DL-based theoretical frameworks, including Reinforcement Learning and Generative Adversarial Networks, specifically for Small Object Detection (SOD) in visible spectral images. The review begins by defining a small object and identifying the datasets available for various applications, such as remote sensing and autonomous vehicles. It then examines the implementation of models according to these datasets and analyzes the findings from other researchers. The analysis reveals that, for most datasets, the average precision (AP) for SOD ranges from 20% to 40% and showcases the need for the advancement and focus.

物体检测在国防、自动驾驶汽车和监视等多个行业都是必不可少的。这些应用依赖于配备摄像头的各种设备，如车辆、无人机和卫星；主要在可见光谱范围内工作，而不是在红外或其他光谱范围内工作。深度学习（DL）技术极大地推动了物体检测领域的发展，使识别各种物体成为可能。然而，探测微小物体仍然是一项具有挑战性的任务。尽管困难重重，但在这些设备捕获的可见光谱图像中识别小物体是至关重要的。探索混合技术和特征架构的修改来解决检测微小物体的挑战是至关重要的。简单的体系结构往往在这方面做得不够，需要更复杂的方法。本文系统地回顾了研究人员以前用来解决这个问题的不同的基于dl的方法。对SOD和DL技术的系统文献综述使用了“系统评价和荟萃分析的首选报告项目”（PRISMA）方法。它讨论了各种基于dl的理论框架，包括强化学习和生成对抗网络，特别是用于可见光谱图像中的小目标检测（SOD）。审查首先定义一个小对象，并确定可用于各种应用的数据集，如遥感和自动驾驶汽车。然后根据这些数据集检查模型的实现，并分析其他研究人员的发现。分析表明，对于大多数数据集，SOD的平均精度（AP）在20%到40%之间，这表明需要改进和关注。

{"title":"A systematic literature review on deep learning approaches for small object detection","authors":"Javed Sayyad, Khush Attarde","doi":"10.1016/j.array.2025.100615","DOIUrl":"10.1016/j.array.2025.100615","url":null,"abstract":"<div><div>Object detection is essential in several industries, including defense, autonomous vehicles, and surveillance. These applications rely on various devices equipped with cameras, such as vehicles, drones, and satellites; primarily operating in the visible spectral domain rather than infrared or other spectral ranges. Deep Learning (DL) techniques have significantly advanced the field of object detection, enabling the identification of various objects. However, detecting tiny objects remains a challenging task. Despite its difficulty, identifying small objects in images captured by these devices in the visible spectrum is crucial. It is essential to explore hybrid techniques and modifications in feature architectures to address the challenge of detecting tiny objects. Simple architectures often fall short in this regard, necessitating more sophisticated approaches. This paper systematically reviews different DL-based approaches researchers have previously employed to tackle this issue. A systematic literature review on SOD and DL techniques uses the ”Preferred Reporting Items for Systematic Reviews and Meta-Analysis” (PRISMA) methodology. It discusses various DL-based theoretical frameworks, including Reinforcement Learning and Generative Adversarial Networks, specifically for Small Object Detection (SOD) in visible spectral images. The review begins by defining a small object and identifying the datasets available for various applications, such as remote sensing and autonomous vehicles. It then examines the implementation of models according to these datasets and analyzes the findings from other researchers. The analysis reveals that, for most datasets, the average precision (AP) for SOD ranges from 20% to 40% and showcases the need for the advancement and focus.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100615"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921199","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

Cognitive load classification during online shopping using deep learning on time series eye movement indices 基于时间序列眼动指数的深度学习在线购物认知负荷分类

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2026-01-03 DOI: 10.1016/j.array.2025.100669

Sunu Wibirama , Muhammad Ainul Fikri , Iman Kahfi Aliza , Kristian Adi Nugraha , Syukron Abu Ishaq Alfarozi , Noor Akhmad Setiawan , Ahmad Riznandi Suhari , Sri Kusrohmaniah

Cognitive load classification during online shopping activities is important to understand the user experience of e-commerce. Traditional classification methods that rely on proprietary software and obtrusive physiological measures often result in inconsistent performance. To address this research gap, we propose a novel approach that leverages deep learning to analyze raw eye movement data during online shopping tasks with low and high cognitive load. The Attention-based Long Short-Term Memory Fully Convolutional Network (ALSTM-FCN) model outperformed other machine learning and deep learning models with an average accuracy and F1 score of 97.70% and 97.69%, respectively. Cognitive load was also measured using the NASA TLX questionnaire, which showed significantly higher scores in high cognitive load tasks for all dimensions: “Mental Demand” (39.37,

p = 0.001

), “Performance” (46.12,

p = 0.004

), “Effort” (51.92,

p = 0.002

), and “Frustration Level” (60.53,

p = 0.001

). Based on the analysis of eye movement features used in cognitive load classification, we found that the variability in eye movement during tasks with low and high cognitive loads was predominantly spatial rather than temporal (

p < 0.05

). Our findings indicate a strong correlation between the deep learning-based classification of raw eye movement data and subjective cognitive load assessments. This study demonstrates the potential of using an affordable eye tracking sensor to classify cognitive load without being constrained by the capability of proprietary software.

网络购物活动中的认知负荷分类对于理解电子商务的用户体验具有重要意义。传统的分类方法依赖于专有软件和突兀的生理测量，结果往往不一致。为了解决这一研究空白，我们提出了一种利用深度学习来分析低认知负荷和高认知负荷在线购物任务期间的原始眼动数据的新方法。基于注意的长短期记忆全卷积网络（ALSTM-FCN）模型的平均准确率和F1分数分别为97.70%和97.69%，优于其他机器学习和深度学习模型。认知负荷也采用NASA TLX问卷进行测量，结果显示，在高认知负荷任务中，“心理需求”（39.37,p=0.001）、“表现”（46.12,p=0.004）、“努力”（51.92,p=0.002）和“沮丧程度”（60.53,p=0.001）的各维度得分均显著较高。基于对认知负荷分类中眼动特征的分析，我们发现低负荷和高负荷任务时眼动的变异性主要是空间变异性而非时间变异性（p<0.05）。我们的研究结果表明，基于深度学习的原始眼动数据分类与主观认知负荷评估之间存在很强的相关性。这项研究证明了使用价格合理的眼动追踪传感器对认知负荷进行分类的潜力，而不受专有软件能力的限制。

{"title":"Cognitive load classification during online shopping using deep learning on time series eye movement indices","authors":"Sunu Wibirama , Muhammad Ainul Fikri , Iman Kahfi Aliza , Kristian Adi Nugraha , Syukron Abu Ishaq Alfarozi , Noor Akhmad Setiawan , Ahmad Riznandi Suhari , Sri Kusrohmaniah","doi":"10.1016/j.array.2025.100669","DOIUrl":"10.1016/j.array.2025.100669","url":null,"abstract":"<div><div>Cognitive load classification during online shopping activities is important to understand the user experience of e-commerce. Traditional classification methods that rely on proprietary software and obtrusive physiological measures often result in inconsistent performance. To address this research gap, we propose a novel approach that leverages deep learning to analyze raw eye movement data during online shopping tasks with low and high cognitive load. The Attention-based Long Short-Term Memory Fully Convolutional Network (ALSTM-FCN) model outperformed other machine learning and deep learning models with an average accuracy and F1 score of 97.70% and 97.69%, respectively. Cognitive load was also measured using the NASA TLX questionnaire, which showed significantly higher scores in high cognitive load tasks for all dimensions: “Mental Demand” (39.37, <span><math><mrow><mi>p</mi><mo>=</mo><mn>0</mn><mo>.</mo><mn>001</mn></mrow></math></span>), “Performance” (46.12, <span><math><mrow><mi>p</mi><mo>=</mo><mn>0</mn><mo>.</mo><mn>004</mn></mrow></math></span>), “Effort” (51.92, <span><math><mrow><mi>p</mi><mo>=</mo><mn>0</mn><mo>.</mo><mn>002</mn></mrow></math></span>), and “Frustration Level” (60.53, <span><math><mrow><mi>p</mi><mo>=</mo><mn>0</mn><mo>.</mo><mn>001</mn></mrow></math></span>). Based on the analysis of eye movement features used in cognitive load classification, we found that the variability in eye movement during tasks with low and high cognitive loads was predominantly spatial rather than temporal (<span><math><mrow><mi>p</mi><mo><</mo><mn>0</mn><mo>.</mo><mn>05</mn></mrow></math></span>). Our findings indicate a strong correlation between the deep learning-based classification of raw eye movement data and subjective cognitive load assessments. This study demonstrates the potential of using an affordable eye tracking sensor to classify cognitive load without being constrained by the capability of proprietary software.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100669"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921320","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

Real-time adaptive neural network-based state of charge prediction of battery pack in a digital twin framework 数字孪生框架下基于实时自适应神经网络的电池组充电状态预测

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2025-12-08 DOI: 10.1016/j.array.2025.100638

Shaik Farooq , M. Harshith , Aneeshsingh Bhatkhande , N. Kumaresan , B. Karthikeyan , A. Rammohan

The accurate and real-time prediction of state of charge (SoC) in battery pack is crucial for the safe and efficient operation of electric vehicles. Traditional estimation methods often suffer from reduced accuracy under sensor errors, battery aging, and dynamic load conditions. This study presents a real-time adaptive neural network (ANN)-based SoC prediction model integrated within a digital twin (DT) framework, designed, and validated using MATLAB/Simulink. The proposed algorithm continuously updates its parameters using real-time current, SoC, and voltage data of battery pack, enabling adaptive learning under varying load and ambient conditions. Compared with traditional methods such as Extended Kalman Filter and particle-filter based estimators, the proposed algorithm reduces the prediction error by 18–22 % and it shortens the response time by 30 %. The simulation results confirm that a strong correlation between the predicted and actual SoC values (R = 0.9999) with a maximum deviation of ±1.5 %. The proposed algorithm demonstrates robust convergence, improved generalization through Bayesian regularization, and high stability during real-time adaptation. This adaptive DT-integrated ANN framework enhances the BMS reliability, supports predictive maintenance, and provides a scalable, and intelligent solution for next-generation electric mobility applications.

准确、实时地预测电池组的荷电状态（SoC）对电动汽车的安全、高效运行至关重要。在传感器误差、电池老化和动态负载条件下，传统的估计方法往往会降低精度。本研究提出了一种集成在数字孪生（DT）框架内的基于实时自适应神经网络（ANN）的SoC预测模型，并使用MATLAB/Simulink进行了设计和验证。该算法利用电池组的实时电流、SoC和电压数据不断更新参数，实现了在不同负载和环境条件下的自适应学习。与传统的扩展卡尔曼滤波和基于粒子滤波的估计方法相比，该算法的预测误差降低了18 - 22%，响应时间缩短了30%。仿真结果表明，预测SoC值与实际SoC值具有较强的相关性（R = 0.9999），最大偏差为±1.5%。该算法具有鲁棒收敛性，通过贝叶斯正则化提高了泛化能力，在实时自适应过程中具有较高的稳定性。这种自适应dt集成ANN框架增强了BMS的可靠性，支持预测性维护，并为下一代电动交通应用提供了可扩展的智能解决方案。

{"title":"Real-time adaptive neural network-based state of charge prediction of battery pack in a digital twin framework","authors":"Shaik Farooq , M. Harshith , Aneeshsingh Bhatkhande , N. Kumaresan , B. Karthikeyan , A. Rammohan","doi":"10.1016/j.array.2025.100638","DOIUrl":"10.1016/j.array.2025.100638","url":null,"abstract":"<div><div>The accurate and real-time prediction of state of charge (SoC) in battery pack is crucial for the safe and efficient operation of electric vehicles. Traditional estimation methods often suffer from reduced accuracy under sensor errors, battery aging, and dynamic load conditions. This study presents a real-time adaptive neural network (ANN)-based SoC prediction model integrated within a digital twin (DT) framework, designed, and validated using MATLAB/Simulink. The proposed algorithm continuously updates its parameters using real-time current, SoC, and voltage data of battery pack, enabling adaptive learning under varying load and ambient conditions. Compared with traditional methods such as Extended Kalman Filter and particle-filter based estimators, the proposed algorithm reduces the prediction error by 18–22 % and it shortens the response time by 30 %. The simulation results confirm that a strong correlation between the predicted and actual SoC values (R = 0.9999) with a maximum deviation of ±1.5 %. The proposed algorithm demonstrates robust convergence, improved generalization through Bayesian regularization, and high stability during real-time adaptation. This adaptive DT-integrated ANN framework enhances the BMS reliability, supports predictive maintenance, and provides a scalable, and intelligent solution for next-generation electric mobility applications.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100638"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145735327","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 parallel particle swarm optimization for improving wireless sensor networks longevity-based dynamic clustering method 基于并行粒子群优化改进无线传感器网络寿命的动态聚类方法

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2025-12-06 DOI: 10.1016/j.array.2025.100633

Ahmed Abdelaziz , Alia Nabil Mahmoud , Vitor Santos

Determining the optimal configuration for wireless sensor networks (WSNs) can be challenging due to the multitude of possible setups. To address this issue, our team has developed the Parallel Particle Swarm Optimization-based Self-Organizing Network Clustering (PPSOPM) method. By taking into account variables like remaining node energy, predictable energy usage, proximity to the base station, and number of nearby nodes, PPSOPM dynamically enhances wireless sensor node clusters. Achieving a balance between these factors is crucial to effectively organize nodes into clusters and select a surrogate node as the cluster's head. In comparison to alternative methods, PPSOPM significantly improves network structure by 44.39 % and extends network lifespan. However, node density may impact network longevity by increasing the distance between nodes. Also, when the base station is far from the sensor area, creating additional clusters can help conserve energy. On average, PPSOPM requires 0.57 s to complete, with a standard deviation of 0.04.

由于可能的设置众多，确定无线传感器网络（wsn）的最佳配置可能具有挑战性。为了解决这个问题，我们的团队开发了基于并行粒子群优化的自组织网络聚类（PPSOPM）方法。通过考虑诸如剩余节点能量、可预测的能源使用、与基站的接近程度以及附近节点的数量等变量，PPSOPM动态增强了无线传感器节点集群。实现这些因素之间的平衡对于有效地将节点组织到集群中并选择代理节点作为集群的头部至关重要。与其他方法相比，PPSOPM显著改善了网络结构44.39%，延长了网络寿命。然而，节点密度可能会增加节点之间的距离，从而影响网络的寿命。此外，当基站远离传感器区域时，创建额外的集群可以帮助节省能源。PPSOPM平均需要0.57 s才能完成，标准差为0.04。

引用次数: 0

Bi-term association based on fuzzy logic 基于模糊逻辑的双项关联

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2026-02-03 DOI: 10.1016/j.array.2026.100704

Wensu Liu , Jing Wan , Na Lv , Xiaobei Zhou

Traditional Boolean search strategies identify co-occurrence but fail to capture semantic relationships. Current relation extraction (RE) frameworks, which rely on domain-specific training, struggle with adaptability to emerging research topics. To address these limitations, we propose a novel framework integrating fuzzy logic with zero-shot large language models (LLMs), enabling the quantification of bi-term associations without explicit training data. Bi-term associations are defined as contextually grounded relationships that extend co-occurrence analysis while avoiding the data annotation requirements of traditional RE. Key contributions include: a fuzzy logic framework for graded membership assignment to capture uncertainty and relationship strength, design of complex fuzzy rules within a lightweight zero-shot LLM, and detailed demonstration of four conceptual case studies in diverse fields, ranging from simple to complex rule design. Benchmark comparisons on RE tasks show that the fuzzy logic method outperforms the baseline model (Qwen3:30b-a3b): F1 scores improve by 0.05 (n2c2 dataset: 0.82 vs. 0.77) and 0.17 (GAD dataset: 0.52 vs. 0.35). However, the fuzzy method is computationally intensive. This work introduces a bi-term association approach based on fuzzy logic to bridge co-occurrence analysis and RE, demonstrating its potential for biomedical knowledge discovery while addressing critical computational challenges in real-world deployment.

传统的布尔搜索策略识别共现，但不能捕获语义关系。当前的关系提取（RE）框架依赖于特定领域的训练，难以适应新兴的研究课题。为了解决这些限制，我们提出了一个新的框架，将模糊逻辑与零射击大语言模型（llm）相结合，使双术语关联的量化没有明确的训练数据。双术语关联被定义为基于上下文的关系，它扩展了共现分析，同时避免了传统正则的数据注释要求。主要贡献包括：用于分级成员分配的模糊逻辑框架，以捕获不确定性和关系强度，在轻量级零shot LLM中设计复杂模糊规则，以及在不同领域（从简单到复杂规则设计）的四个概念案例研究的详细演示。在RE任务上的基准比较表明，模糊逻辑方法优于基线模型（Qwen3:30b-a3b）： F1得分提高0.05 （n2c2数据集：0.82 vs. 0.77）和0.17 （GAD数据集：0.52 vs. 0.35）。然而，模糊方法的计算量很大。这项工作介绍了一种基于模糊逻辑的双术语关联方法，以桥接共现分析和RE，展示了其在解决现实世界部署中的关键计算挑战的同时，在生物医学知识发现方面的潜力。

{"title":"Bi-term association based on fuzzy logic","authors":"Wensu Liu , Jing Wan , Na Lv , Xiaobei Zhou","doi":"10.1016/j.array.2026.100704","DOIUrl":"10.1016/j.array.2026.100704","url":null,"abstract":"<div><div>Traditional Boolean search strategies identify co-occurrence but fail to capture semantic relationships. Current relation extraction (RE) frameworks, which rely on domain-specific training, struggle with adaptability to emerging research topics. To address these limitations, we propose a novel framework integrating fuzzy logic with zero-shot large language models (LLMs), enabling the quantification of bi-term associations without explicit training data. Bi-term associations are defined as contextually grounded relationships that extend co-occurrence analysis while avoiding the data annotation requirements of traditional RE. Key contributions include: a fuzzy logic framework for graded membership assignment to capture uncertainty and relationship strength, design of complex fuzzy rules within a lightweight zero-shot LLM, and detailed demonstration of four conceptual case studies in diverse fields, ranging from simple to complex rule design. Benchmark comparisons on RE tasks show that the fuzzy logic method outperforms the baseline model (Qwen3:30b-a3b): F1 scores improve by 0.05 (n2c2 dataset: 0.82 vs. 0.77) and 0.17 (GAD dataset: 0.52 vs. 0.35). However, the fuzzy method is computationally intensive. This work introduces a bi-term association approach based on fuzzy logic to bridge co-occurrence analysis and RE, demonstrating its potential for biomedical knowledge discovery while addressing critical computational challenges in real-world deployment.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100704"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146184899","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

Federated Convolutional Neural Networks (F-CNNs) for privacy-preserving multi-class skin lesion classification 联邦卷积神经网络（f - cnn）用于保护隐私的多类皮肤病变分类

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2025-12-30 DOI: 10.1016/j.array.2025.100667

Khadija Shahzad , Anum Khashir , Hina Tufail , Abdul Ahad , Zahra Ali , Filipe Madeira , Ivan Miguel Pires

Skin lesions include a variety of abnormalities found on the skin. These may be benign (not cancerous) or malignant (cancerous). Every year, the number of cases of skin cancer increases globally, increasing the death rate. Medical data is scarce because people are reluctant to provide their health information due to privacy concerns. In this research, a decentralized machine learning approach, Federated learning, is the primary focus of the discipline to preserve patient data. Using this method, models are trained independently on several dispersed devices without sharing the data. To balance the data and enrich the dataset, the Synthetic Minority Over-sampling technique with Edited Nearest Neighbors (SMOTEENN) is used in this study. The HAM10000 dataset was benchmarked using a Convolutional Neural Network (CNN). Seven classes of HAM10000 include vascular skin lesions, benign keratosis, actinic keratosis, melanoma, dermatofibroma, and melanocytic nevi. A centralized method yields an accuracy of 99.39%, and f1-score, precision, and recall of 99.00%. A simulated Federated learning with three clients, ten rounds, and thirty training epochs produced 93.00% precision, 92.00% recall, 92.00% f1-score, and 91.80% accuracy, respectively. At the same time, an increase to four clients and thirty training epochs produced an accuracy, recall, precision, and f1-score of 97.00% with ten rounds.

皮肤病变包括在皮肤上发现的各种异常。这些可能是良性的（非癌性的）或恶性的（癌性的）。每年，全球皮肤癌病例数都在增加，死亡率也在增加。医疗数据很少，因为人们出于隐私考虑不愿提供自己的健康信息。在这项研究中，分散的机器学习方法，联邦学习，是该学科保存患者数据的主要焦点。使用该方法，模型在多个分散的设备上独立训练，而不共享数据。为了平衡数据和丰富数据集，本研究使用了编辑近邻的合成少数派过采样技术（SMOTEENN）。HAM10000数据集使用卷积神经网络（CNN）进行基准测试。HAM10000分为7类：血管性皮肤病变、良性角化病、光化性角化病、黑色素瘤、皮肤纤维瘤、黑素细胞痣。集中式方法的准确率为99.39%，f1-score、精密度和召回率为99.00%。模拟的联邦学习有3个客户端、10轮和30个训练周期，分别产生了93.00%的准确率、92.00%的召回率、92.00%的f1得分和91.80%的准确率。与此同时，增加到4个客户和30个训练周期，在10轮训练中，准确率、召回率、精确度和f1得分达到97.00%。

{"title":"Federated Convolutional Neural Networks (F-CNNs) for privacy-preserving multi-class skin lesion classification","authors":"Khadija Shahzad , Anum Khashir , Hina Tufail , Abdul Ahad , Zahra Ali , Filipe Madeira , Ivan Miguel Pires","doi":"10.1016/j.array.2025.100667","DOIUrl":"10.1016/j.array.2025.100667","url":null,"abstract":"<div><div>Skin lesions include a variety of abnormalities found on the skin. These may be benign (not cancerous) or malignant (cancerous). Every year, the number of cases of skin cancer increases globally, increasing the death rate. Medical data is scarce because people are reluctant to provide their health information due to privacy concerns. In this research, a decentralized machine learning approach, Federated learning, is the primary focus of the discipline to preserve patient data. Using this method, models are trained independently on several dispersed devices without sharing the data. To balance the data and enrich the dataset, the Synthetic Minority Over-sampling technique with Edited Nearest Neighbors (SMOTEENN) is used in this study. The HAM10000 dataset was benchmarked using a Convolutional Neural Network (CNN). Seven classes of HAM10000 include vascular skin lesions, benign keratosis, actinic keratosis, melanoma, dermatofibroma, and melanocytic nevi. A centralized method yields an accuracy of 99.39%, and f1-score, precision, and recall of 99.00%. A simulated Federated learning with three clients, ten rounds, and thirty training epochs produced 93.00% precision, 92.00% recall, 92.00% f1-score, and 91.80% accuracy, respectively. At the same time, an increase to four clients and thirty training epochs produced an accuracy, recall, precision, and f1-score of 97.00% with ten rounds.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100667"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973363","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

Integrating feature fusion with hybrid optimization for multiple sclerosis MRI classification 融合特征融合与混合优化的多发性硬化MRI分类

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2026-01-06 DOI: 10.1016/j.array.2025.100677

Nandini Anam , Sharief Basha S , Chiranji Lal Chowdhary

Detecting Multiple Sclerosis (MS) has previously been difficult to identify with MRI scans due to the subtlety and dispersion of lesions, as well as other imaging anomalies. The researchers in this study created a unique hybrid framework that combines two state-of-the-art convolutional neural networks, ResNet-50 and EfficientNet-B7. Additionally, a new hybrid bio-inspired optimization strategy combining the Grey Wolf Optimizer (GWO) and the Genetic Algorithm (GA) is described. The technique simplifies the computations and ensures that the best characteristics are picked. We extracted deep features from both CNNs, used Principal Component Analysis (PCA) to reduce them to high dimensions, and then employed the GA-GWO method to identify suitable features. The enhanced artificial neural network (ANN) classifier outperformed standalone CNN-based models, achieving a maximum accuracy of 90.67 %(refer to table 4). The suggested framework for dependable and comprehensible precision and processing efficacy. This work has the potential to motivate future studies in related areas.

由于病变的微妙性和弥散性以及其他成像异常，以前很难通过MRI扫描来识别多发性硬化症（MS）。在这项研究中，研究人员创建了一个独特的混合框架，结合了两个最先进的卷积神经网络ResNet-50和EfficientNet-B7。此外，还提出了一种将灰狼优化器（GWO）与遗传算法（GA）相结合的混合生物优化策略。该技术简化了计算，并确保了最佳特征的选择。我们从两个cnn中提取深度特征，使用主成分分析（PCA）将其降维到高维，然后使用GA-GWO方法识别合适的特征。增强的人工神经网络（ANN）分类器优于独立的基于cnn的模型，最高准确率达到90.67%（见表4）。建议的框架是可靠和可理解的精度和加工效率。这项工作有可能激发未来相关领域的研究。

引用次数: 0

Ret-UNet: Enhancing medical image segmentation with self-retention Ret-UNet：增强医学图像的自保留分割

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2025-12-18 DOI: 10.1016/j.array.2025.100653

Tianjun Guo, Weixin Zhao, Jian Peng

Medical image segmentation has advanced significantly due to the integration of deep learning techniques, particularly convolutional neural networks (CNNs) like U-Net. However, CNNs often struggle to capture global spatial relationships, which are crucial for accurately segmenting complex anatomical structures. To address this limitation, we propose Ret-UNet, a novel architecture that enhances the traditional U-Net framework by incorporating the Self-Retention mechanism. Self-Retention introduces an explicit shape prior related to the Euclidean distance, which effectively encode global spatial relationships within the image. The Ret-UNet leverages both local feature extraction and global context awareness by incorporating Ret Blocks into the U-Net like architecture, leading to improved segmentation performance. Evaluations on ACDC, CAMUS and Synapse datasets demonstrate that Ret-UNet achieves superior segmentation accuracy and robustness, outperforming state-of-the-art models. The code is available at https://github.com/weirdgit/RetUNet.

由于深度学习技术的集成，特别是卷积神经网络（cnn），如U-Net，医学图像分割取得了显著进展。然而，cnn经常难以捕捉全局空间关系，这对于准确分割复杂的解剖结构至关重要。为了解决这一限制，我们提出了Ret-UNet，这是一种新的架构，通过结合自保留机制来增强传统的U-Net框架。自保留引入了与欧几里得距离相关的显式形状先验，有效地编码了图像中的全局空间关系。Ret- unet通过将Ret块合并到类似U-Net的架构中，利用了局部特征提取和全局上下文感知，从而提高了分割性能。对ACDC、CAMUS和Synapse数据集的评估表明，Ret-UNet实现了卓越的分割精度和鲁棒性，优于最先进的模型。代码可在https://github.com/weirdgit/RetUNet上获得。

引用次数: 0

Exploring and identifying fine-grained accessibility issues in app store using fine-tuned deep learning 使用微调深度学习探索和识别应用商店中的细粒度可访问性问题

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2025-12-11 DOI: 10.1016/j.array.2025.100572

Mumrez Khan , Zhixiao Wang , Javed Ali Khan , Nek Dil Khan

The Apple App Store (AAS) allows users to provide feedback on applications, offering developers insights into improving software performance. Researchers have utilized this feedback for software evolution activities, including features, issues, and nonfunctional requirements. However, end-user feedback has not been explored to identify accessibility-related challenges. This study proposes an automated approach to detect and classify accessibility issues by analyzing end-user reviews in the AAS. We crawled 178667 user reviews from 85 apps across 18 categories to represent a diverse sample. We developed a coding guideline to identify common accessibility issues, including Navigation and Interaction Problems (NAV), Input and Control Issues (INPUT), Compatibility with Assistive Technologies (CAT), Audio and visual accessibility issues (AUDIOVISUAL), and UI Accessibility Issues (UI). We manually annotated reviews using coding guidelines and content analysis to create a labeled dataset for training and evaluating deep learning(DL) algorithms to detect accessibility in user comments and classify them into categories. The experiments showed that fine-tuned DL classifiers achieved high accuracy in detecting accessibility and classifying them into specific types. For binary classification, the CNN classifier achieved 93% precision, while LSTM, BiLSTM, GRU, and BiGRU achieved accuracies from 76% to 87%. In fine-grained classification, CNN performed better with 97% accuracy, followed by BiGRU and BiLSTM at 96%. The BiLSTM and LSTM models demonstrated strong performance, with accuracies of 96% and 95%. These results show the potential of automated methods to improve identification of accessibility challenges, helping developers address these issues effectively and enhance user experience.

苹果应用商店（AAS）允许用户对应用程序提供反馈，为开发人员提供改进软件性能的见解。研究人员已经将这种反馈用于软件进化活动，包括特性、问题和非功能需求。然而，最终用户的反馈还没有被用来确定可访问性相关的挑战。本研究提出了一种自动化的方法，通过分析AAS中的最终用户评论来检测和分类可访问性问题。我们从18个类别的85款应用中抓取了178667条用户评论。我们开发了一个编码指南来识别常见的可访问性问题，包括导航和交互问题（NAV）、输入和控制问题（Input）、与辅助技术的兼容性（CAT）、音频和视觉可访问性问题（AUDIOVISUAL）和UI可访问性问题（UI）。我们使用编码指南和内容分析手动注释评论，以创建标记数据集，用于训练和评估深度学习（DL）算法，以检测用户评论中的可访问性并将其分类。实验表明，经过微调的深度学习分类器在检测可达性并将其分类为特定类型方面具有较高的准确率。对于二值分类，CNN分类器的准确率达到93%，而LSTM、BiLSTM、GRU和BiGRU的准确率在76%到87%之间。在细粒度分类中，CNN表现较好，准确率为97%，其次是BiGRU和BiLSTM，准确率为96%。BiLSTM和LSTM模型表现出较强的性能，准确率分别为96%和95%。这些结果显示了自动化方法的潜力，可以改进易访问性挑战的识别，帮助开发人员有效地解决这些问题并增强用户体验。

{"title":"Exploring and identifying fine-grained accessibility issues in app store using fine-tuned deep learning","authors":"Mumrez Khan , Zhixiao Wang , Javed Ali Khan , Nek Dil Khan","doi":"10.1016/j.array.2025.100572","DOIUrl":"10.1016/j.array.2025.100572","url":null,"abstract":"<div><div>The Apple App Store (AAS) allows users to provide feedback on applications, offering developers insights into improving software performance. Researchers have utilized this feedback for software evolution activities, including features, issues, and nonfunctional requirements. However, end-user feedback has not been explored to identify accessibility-related challenges. This study proposes an automated approach to detect and classify accessibility issues by analyzing end-user reviews in the AAS. We crawled 178667 user reviews from 85 apps across 18 categories to represent a diverse sample. We developed a coding guideline to identify common accessibility issues, including Navigation and Interaction Problems (NAV), Input and Control Issues (INPUT), Compatibility with Assistive Technologies (CAT), Audio and visual accessibility issues (AUDIOVISUAL), and UI Accessibility Issues (UI). We manually annotated reviews using coding guidelines and content analysis to create a labeled dataset for training and evaluating deep learning(DL) algorithms to detect accessibility in user comments and classify them into categories. The experiments showed that fine-tuned DL classifiers achieved high accuracy in detecting accessibility and classifying them into specific types. For binary classification, the CNN classifier achieved 93% precision, while LSTM, BiLSTM, GRU, and BiGRU achieved accuracies from 76% to 87%. In fine-grained classification, CNN performed better with 97% accuracy, followed by BiGRU and BiLSTM at 96%. The BiLSTM and LSTM models demonstrated strong performance, with accuracies of 96% and 95%. These results show the potential of automated methods to improve identification of accessibility challenges, helping developers address these issues effectively and enhance user experience.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100572"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788379","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

An automated multi-scale and multi-contextual MobileNetv3 for malware detection based on IoT 基于物联网的自动多尺度和多上下文MobileNetv3恶意软件检测

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2026-03-01 Epub Date: 2026-01-20 DOI: 10.1016/j.array.2026.100681

Sidra Javed , Guowei Wu , Hamza Javed , Osama A. Khashan , Haseeb Hassan , Anwar Ghani

Malware detection is a crucial aspect of cybersecurity, aimed at identifying and mitigating malicious software that poses threats to systems and networks. Traditional malware detection methods face challenges in terms of both detection accuracy and computational cost, as deep learning models can be resource-intensive and difficult to deploy in real-time environments. This paper introduces the novel MSMC-MobileNet (Multi-Scale and Multi-Contextual MobileNet) malware detection and classification model, designed to address the challenges of accuracy and computational cost. First, MobileNetv3 is used to extract features from the dataset. To enhance feature extraction, the SE (Squeeze-and-Excitation) module is integrated, focusing on the region of interest using an attention mechanism. The multiscale and multicontextual features are extracted using the ASPP (Atrous Spatial Pyramid Pooling) and FPP (Feature Pyramid Pooling) modules. Channel-wise pruning is applied to the ASPP and FPP modules, reducing computational cost. The model is evaluated on the publicly available Malimg and MaleVis datasets. The proposed MSMC-MobileNet model achieves impressive performance with 92.37% accuracy, 96.54% precision, 95.84% recall, 95.47% F1 score, and 98.59% AUC on the Malimg dataset. On the MaleVis dataset, the model yields 95.08% accuracy, 98.33% precision, 97.9% recall, 98.15% F1 score, and 96.98% AUC. When both datasets are combined, the MSMC-MobileNet achieves 98.79% accuracy, 99.84% precision, 99.73% recall, 99.89% F1 score, and 1.00 AUC. Despite its high accuracy, the model remains computationally efficient, outperforming state-of-the-art methods in both detection performance and computational cost.

恶意软件检测是网络安全的一个重要方面，旨在识别和减轻对系统和网络构成威胁的恶意软件。传统的恶意软件检测方法在检测精度和计算成本方面都面临挑战，因为深度学习模型可能是资源密集型的，并且难以在实时环境中部署。本文介绍了一种新的MSMC-MobileNet （Multi-Scale and Multi-Contextual MobileNet）恶意软件检测与分类模型，旨在解决准确性和计算成本方面的挑战。首先，使用MobileNetv3从数据集中提取特征。为了增强特征提取，集成了SE （Squeeze-and-Excitation）模块，使用注意机制聚焦于感兴趣的区域。利用空间金字塔池（ASPP）和特征金字塔池（FPP）模块提取多尺度和多上下文特征。在ASPP和FPP模块中应用了通道方向剪枝，减少了计算成本。该模型在公开可用的Malimg和MaleVis数据集上进行了评估。本文提出的MSMC-MobileNet模型在Malimg数据集上的准确率为92.37%，精密度为96.54%，召回率为95.84%，F1分数为95.47%，AUC为98.59%。在MaleVis数据集上，该模型的准确率为95.08%，精度为98.33%，召回率为97.9%，F1得分为98.15%，AUC为96.98%。当两个数据集结合在一起时，MSMC-MobileNet的准确率为98.79%，精密度为99.84%，召回率为99.73%，F1分数为99.89%，AUC为1.00。尽管该模型具有很高的精度，但其计算效率仍然很高，在检测性能和计算成本方面都优于最先进的方法。

{"title":"An automated multi-scale and multi-contextual MobileNetv3 for malware detection based on IoT","authors":"Sidra Javed , Guowei Wu , Hamza Javed , Osama A. Khashan , Haseeb Hassan , Anwar Ghani","doi":"10.1016/j.array.2026.100681","DOIUrl":"10.1016/j.array.2026.100681","url":null,"abstract":"<div><div>Malware detection is a crucial aspect of cybersecurity, aimed at identifying and mitigating malicious software that poses threats to systems and networks. Traditional malware detection methods face challenges in terms of both detection accuracy and computational cost, as deep learning models can be resource-intensive and difficult to deploy in real-time environments. This paper introduces the novel MSMC-MobileNet (Multi-Scale and Multi-Contextual MobileNet) malware detection and classification model, designed to address the challenges of accuracy and computational cost. First, MobileNetv3 is used to extract features from the dataset. To enhance feature extraction, the SE (Squeeze-and-Excitation) module is integrated, focusing on the region of interest using an attention mechanism. The multiscale and multicontextual features are extracted using the ASPP (Atrous Spatial Pyramid Pooling) and FPP (Feature Pyramid Pooling) modules. Channel-wise pruning is applied to the ASPP and FPP modules, reducing computational cost. The model is evaluated on the publicly available Malimg and MaleVis datasets. The proposed MSMC-MobileNet model achieves impressive performance with 92.37% accuracy, 96.54% precision, 95.84% recall, 95.47% F1 score, and 98.59% AUC on the Malimg dataset. On the MaleVis dataset, the model yields 95.08% accuracy, 98.33% precision, 97.9% recall, 98.15% F1 score, and 96.98% AUC. When both datasets are combined, the MSMC-MobileNet achieves 98.79% accuracy, 99.84% precision, 99.73% recall, 99.89% F1 score, and 1.00 AUC. Despite its high accuracy, the model remains computationally efficient, outperforming state-of-the-art methods in both detection performance and computational cost.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100681"},"PeriodicalIF":4.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034470","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

Array最新文献