IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-19 DOI: 10.1016/j.array.2025.100651

Sanchari Ghosh , Sandeep Mondal , Nishit Kumar Srivastava

Digital Platform Services Supply Chains (DPSSC) form the operational backbone of contemporary digital economies, yet their sustainability performance is shaped by interdependent technical, organisational and behavioural constraints that remain insufficiently examined. This study addresses this gap by developing unified framework based on DEMATEL-ANP (DANP) and NSGA-II to identify, prioritise and test 7 sustainability barriers spanning from B1-B7. Expert elicitation and DANP analyses reveal a stable causal structure, in which three upstream systemic barriers B1 (ESG standardisation for digital platforms), B3 (energy-intensive infrastructure) and B2 (sustainable tech-innovation by industry-academia collaboration) emerge as system drivers with together accounting for nearly 70 % of global priority weight, whereas B4 (curricular gaps), B5 (behavioural nudges), B6 (algorithmic ESG indicators), B7 (carbon accountability) function as dependent barriers with limited leverage in the absence of upstream correction. NSGA-II is applied to quantify trade-offs between emission reduction and implementation effort under a parameterised case setting informed by publicly disclosed sustainability data. The optimisation yields stable Pareto fronts across convergence, hypervolume, spacing, multi-seed and uncertainty diagnostics. Strategies emphasising B1-B3 deliver largest marginal mitigation benefits within the efficiency zone, while interventions centred on B4-B7 rapidly encounter diminishing returns. This integrated evidence demonstrates that DPSSC sustainability follows a two-stage intervention logic one by addressing upstream structural enablers, the other followed by scaling downstream behavioural, algorithmic and logistical measures once system-level constraints are resolved. Thus, the study provides a transparent and decision-relevant basis for prioritising sustainability actions, strengthening industry-academia engagement and aligning digital platform operations with United Nations SDG 4,7,9,11–13 targets.

数字平台服务供应链（DPSSC）是当代数字经济的运营支柱，但其可持续性表现受到相互依赖的技术、组织和行为约束的影响，而这些制约因素仍未得到充分研究。本研究通过开发基于DEMATEL-ANP （DANP）和NSGA-II的统一框架来识别、优先考虑和测试从b1到b7的7个可持续性障碍，从而解决了这一差距。专家启发和DANP分析揭示了一个稳定的因果结构，其中三个上游系统性障碍B1（数字平台的ESG标准化），B3（能源密集型基础设施）和B2（产学研合作的可持续技术创新）作为系统驱动因素出现，合计占全球优先权重的近70%，而B4（课程差距），B5（行为推动），B6（算法ESG指标），B7（碳问责制）在缺乏上游纠正的情况下作为依赖障碍发挥有限的杠杆作用。NSGA-II应用于在公开披露的可持续性数据的参数化案例设置下量化减排和实施努力之间的权衡。优化产生稳定的帕累托战线跨越收敛，超大容量，间距，多种子和不确定性诊断。强调B1-B3的战略在效率区内产生最大的边际缓解效益，而以B4-B7为中心的干预措施的收益迅速递减。这些综合证据表明，DPSSC的可持续性遵循两个阶段的干预逻辑，一个是解决上游结构因素，另一个是解决系统级约束后扩展下游行为、算法和后勤措施。因此，该研究为确定可持续发展行动的优先顺序、加强产学研合作以及将数字平台运营与联合国可持续发展目标4、7、9、11-13相一致提供了透明和决策相关的基础。

{"title":"Evaluating sustainability barriers for digital platform services supply chain: A study on strategic intervention through industry-academia collaboration","authors":"Sanchari Ghosh , Sandeep Mondal , Nishit Kumar Srivastava","doi":"10.1016/j.array.2025.100651","DOIUrl":"10.1016/j.array.2025.100651","url":null,"abstract":"<div><div>Digital Platform Services Supply Chains (DPSSC) form the operational backbone of contemporary digital economies, yet their sustainability performance is shaped by interdependent technical, organisational and behavioural constraints that remain insufficiently examined. This study addresses this gap by developing unified framework based on DEMATEL-ANP (DANP) and NSGA-II to identify, prioritise and test 7 sustainability barriers spanning from B1-B7. Expert elicitation and DANP analyses reveal a stable causal structure, in which three upstream systemic barriers B1 (ESG standardisation for digital platforms), B3 (energy-intensive infrastructure) and B2 (sustainable tech-innovation by industry-academia collaboration) emerge as system drivers with together accounting for nearly 70 % of global priority weight, whereas B4 (curricular gaps), B5 (behavioural nudges), B6 (algorithmic ESG indicators), B7 (carbon accountability) function as dependent barriers with limited leverage in the absence of upstream correction. NSGA-II is applied to quantify trade-offs between emission reduction and implementation effort under a parameterised case setting informed by publicly disclosed sustainability data. The optimisation yields stable Pareto fronts across convergence, hypervolume, spacing, multi-seed and uncertainty diagnostics. Strategies emphasising B1-B3 deliver largest marginal mitigation benefits within the efficiency zone, while interventions centred on B4-B7 rapidly encounter diminishing returns. This integrated evidence demonstrates that DPSSC sustainability follows a two-stage intervention logic one by addressing upstream structural enablers, the other followed by scaling downstream behavioural, algorithmic and logistical measures once system-level constraints are resolved. Thus, the study provides a transparent and decision-relevant basis for prioritising sustainability actions, strengthening industry-academia engagement and aligning digital platform operations with United Nations SDG 4,7,9,11–13 targets.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100651"},"PeriodicalIF":4.5,"publicationDate":"2025-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921781","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

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 : 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

Secure access to implantable medical devices: A deep learning-driven protocol using ECG signals 安全访问植入式医疗设备：使用ECG信号的深度学习驱动协议

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-17 DOI: 10.1016/j.array.2025.100650

Amirhossein Safari , Mohsen Hooshmand , Sadegh Sadeghi , Peyman Phalevani , Nasour Bagheri , Mahmoud Jabbarpour

The Internet of Things in medical applications demands strong privacy and security, particularly for implantable medical devices (IMDs) that must be reconfigured wirelessly through programmer devices. Existing IMD access-control protocols offer partial protection but still face challenges in accuracy and reliability. Biometric signals, especially Electrocardiogram (ECG), provide a promising alternative for secure and continuous authentication in such IoT systems. This work proposes four deep learning-based authorization approaches enabling secure communication between an IMD and a nearby programmer using ECG-derived representations. The methods fall into two categories: (1) Reliable IMD (RI) and unReliable IMD (uRI), where RI assumes the IMD’s internally recorded ECG is trustworthy (authentication uses only the programmer’s signal), while uRI requires cross-verification between IMD and programmer signals; and (2) Omni (multi-user authentication) versus Mono (single-user authentication). Mono-RI and Omni-RI employ convolutional neural network (CNN)-based architectures, whereas Mono-uRI and Omni-uRI use Siamese networks to support both authentication and replay attack detection. Experimental results show that all four models outperform state-of-the-art solutions. Mono-RI achieves 97.73% authentication accuracy, while Omni-uRI reaches 98.75%. Replay attack detection rates are 99.19% for Mono models and 90.73% for Omni models. These findings demonstrate that the proposed methods are effective across different operational scenarios, supporting both high-security modes and emergency modes that demand rapid access, depending on the desired balance between performance and computational complexity.

医疗应用中的物联网需要强大的隐私和安全性，特别是对于必须通过编程设备无线重新配置的植入式医疗设备（imd）。现有的IMD访问控制协议提供了部分保护，但在准确性和可靠性方面仍然面临挑战。生物识别信号，特别是心电图（ECG），为此类物联网系统中的安全和连续认证提供了一种有前途的替代方案。这项工作提出了四种基于深度学习的授权方法，可以使用ecg衍生的表示在IMD和附近的程序员之间实现安全通信。这些方法分为两类：(1)可靠IMD （RI）和不可靠IMD (uRI)，其中RI假设IMD内部记录的ECG是可信的（认证仅使用程序员的信号），而uRI需要在IMD和程序员信号之间进行交叉验证；(2) Omni（多用户身份验证）与Mono（单用户身份验证）。Mono-RI和Omni-RI采用基于卷积神经网络（CNN）的架构，而Mono-uRI和Omni-uRI使用暹罗网络来支持身份验证和重放攻击检测。实验结果表明，所有四种模型都优于最先进的解决方案。Mono-RI认证准确率为97.73%，Omni-uRI认证准确率为98.75%。Mono模型的重放攻击检测率为99.19%，Omni模型为90.73%。这些发现表明，所提出的方法在不同的操作场景中都是有效的，既支持高安全性模式，也支持需要快速访问的紧急模式，这取决于性能和计算复杂性之间的理想平衡。

{"title":"Secure access to implantable medical devices: A deep learning-driven protocol using ECG signals","authors":"Amirhossein Safari , Mohsen Hooshmand , Sadegh Sadeghi , Peyman Phalevani , Nasour Bagheri , Mahmoud Jabbarpour","doi":"10.1016/j.array.2025.100650","DOIUrl":"10.1016/j.array.2025.100650","url":null,"abstract":"<div><div>The Internet of Things in medical applications demands strong privacy and security, particularly for implantable medical devices (IMDs) that must be reconfigured wirelessly through programmer devices. Existing IMD access-control protocols offer partial protection but still face challenges in accuracy and reliability. Biometric signals, especially Electrocardiogram (ECG), provide a promising alternative for secure and continuous authentication in such IoT systems. This work proposes four deep learning-based authorization approaches enabling secure communication between an IMD and a nearby programmer using ECG-derived representations. The methods fall into two categories: (1) Reliable IMD (RI) and unReliable IMD (uRI), where RI assumes the IMD’s internally recorded ECG is trustworthy (authentication uses only the programmer’s signal), while uRI requires cross-verification between IMD and programmer signals; and (2) Omni (multi-user authentication) versus Mono (single-user authentication). Mono-RI and Omni-RI employ convolutional neural network (CNN)-based architectures, whereas Mono-uRI and Omni-uRI use Siamese networks to support both authentication and replay attack detection. Experimental results show that all four models outperform state-of-the-art solutions. Mono-RI achieves 97.73% authentication accuracy, while Omni-uRI reaches 98.75%. Replay attack detection rates are 99.19% for Mono models and 90.73% for Omni models. These findings demonstrate that the proposed methods are effective across different operational scenarios, supporting both high-security modes and emergency modes that demand rapid access, depending on the desired balance between performance and computational complexity.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100650"},"PeriodicalIF":4.5,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788322","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

HitHire: The future of ethical, fair, and sustainable AI recruitment – A governance framework HitHire：道德、公平和可持续的人工智能招聘的未来——一个治理框架

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-17 DOI: 10.1016/j.array.2025.100592

Elham Albaroudi , Taha Mansouri , Mohammad Hatamleh , Ali Alameer

Artificial Intelligence (AI) is transforming recruitment but remains susceptible to algorithmic bias and environmental inefficiencies. This paper presents HitHire, a pilot fairness- and sustainability-aware AI hiring platform tailored to the Saudi Arabian context and aligned with Vision 2030 goals. HitHire integrates large language models (LLMs), adversarial debiasing, Shapley Additive Explanations (SHAP), and real-time carbon tracking to ensure transparent and equitable candidate ranking. Evaluated on 350 anonymized CVs across four job roles (web development, finance, human resources, and data science) using a 70/20/10 train/test/validation split, HitHire achieves notable improvements in fairness metrics—Statistical Parity Difference (SPD) for gender = 0.0156 and Disparate Impact (DI) for nationality = 1.2387—while maintaining strong predictive performance (F1 = 0.96 compared to a baseline of 0.80). The system achieves over a 40% reduction in operational CO

_{2}

emissions, with inference energy consumption of 0.003 kWh per query. In a three-month pilot study involving 23 HR professionals within a large Saudi organization, 87% of participants rated system trust at 4 out of 5 or higher. These findings contribute to national digital ethics strategies such as the Saudi Green Initiative, which emphasizes carbon neutrality and sustainable innovation.

人工智能（AI）正在改变招聘，但仍然容易受到算法偏见和环境效率低下的影响。本文介绍了HitHire，这是一个针对沙特阿拉伯国情量身定制的具有公平和可持续性意识的人工智能招聘平台，与2030年愿景目标保持一致。HitHire集成了大型语言模型（LLMs）、对抗式去偏见、Shapley Additive Explanations （SHAP）和实时碳跟踪，以确保透明和公平的候选人排名。HitHire使用70/20/10训练/测试/验证分割法对四种工作角色（网络开发、金融、人力资源和数据科学）的350份匿名简历进行评估，在公平性指标上取得了显著改善——性别的统计均等差异(SPD) = 0.0156，国籍的差异影响(DI) = 1.2387——同时保持了强大的预测性能（F1 = 0.96，而基线为0.80）。该系统减少了40%以上的运行二氧化碳排放量，每次查询的推断能耗为0.003千瓦时。在一项为期三个月的试点研究中，一家大型沙特组织的23名人力资源专业人士参与了这项研究，87%的参与者将系统信任度评为4分（满分5分）或更高。这些发现有助于制定国家数字道德战略，如强调碳中和和可持续创新的沙特绿色倡议。

{"title":"HitHire: The future of ethical, fair, and sustainable AI recruitment – A governance framework","authors":"Elham Albaroudi , Taha Mansouri , Mohammad Hatamleh , Ali Alameer","doi":"10.1016/j.array.2025.100592","DOIUrl":"10.1016/j.array.2025.100592","url":null,"abstract":"<div><div>Artificial Intelligence (AI) is transforming recruitment but remains susceptible to algorithmic bias and environmental inefficiencies. This paper presents HitHire, a pilot fairness- and sustainability-aware AI hiring platform tailored to the Saudi Arabian context and aligned with Vision 2030 goals. HitHire integrates large language models (LLMs), adversarial debiasing, Shapley Additive Explanations (SHAP), and real-time carbon tracking to ensure transparent and equitable candidate ranking. Evaluated on 350 anonymized CVs across four job roles (web development, finance, human resources, and data science) using a 70/20/10 train/test/validation split, HitHire achieves notable improvements in fairness metrics—Statistical Parity Difference (SPD) for gender = 0.0156 and Disparate Impact (DI) for nationality = 1.2387—while maintaining strong predictive performance (F1 = 0.96 compared to a baseline of 0.80). The system achieves over a 40% reduction in operational CO<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span> emissions, with inference energy consumption of 0.003 kWh per query. In a three-month pilot study involving 23 HR professionals within a large Saudi organization, 87% of participants rated system trust at 4 out of 5 or higher. These findings contribute to national digital ethics strategies such as the Saudi Green Initiative, which emphasizes carbon neutrality and sustainable innovation.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100592"},"PeriodicalIF":4.5,"publicationDate":"2025-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788016","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

Enhancing resilience through bilinear attention pooled region proposal network for metal sheet grid defect classification 利用双线性注意力池区域建议网络增强金属薄板网格缺陷分类的弹性

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-16 DOI: 10.1016/j.array.2025.100647

Shyamala Devi M , Priya S , Yuvaraj Natarajan , Sri Preethaa K.R

Defect detection in metal sheet grids is a critical task in the construction industry, where structural reliability and aesthetic quality are crucial. This paper proposes Bilinear Attention Pooled Faster Region-based Convolutional Neural Network (BF-RCNN) to identify and classify surface defects with high accuracy to enhance the building integrity. The novelty of BF-RCNN model relies on generation of Neuro fuzzy weighted otsu (NFO) from 2,400 metal sheet grid images from the public repository. By combining the idea of fuzzy logic and adaptive thresholding, the visibility of defective regions is enhanced by NFO filtering. CHS images are processed by BF-RCNN that uses ResNet-50 for deep multi-scale feature extraction. To strengthen the model's attention to informative features, a Squeeze-and-Excitation (SE) block is integrated into BF-RCNN, enabling channel-wise recalibration and improved sensitivity to subtle defect patterns. Further, Compact Bilinear Pooling (CBP) is employed to model second-order feature interactions efficiently. Also, the representation of complex textures and defect anomalies are enhanced without increasing computational cost. Experimental evaluations show that proposed BF-RCNN model achieves high accuracy, precision and recall of 99.82, 99.12 and 98.94 respectively against existing CNN models. The proposed BF-RCNN model offers significant practical benefits for real-world manufacturing and quality control processes. It also enables high reliable identification of surface defects in metal sheet grids. Thus, the model reduces manual inspection time, minimizes human error, and ensures consistent product quality. This BF-RCNN presents a scalable and intelligent solution for quality inspection in construction industry with both accuracy and reliability in defect assessment processes.

在建筑行业中，金属板网格的缺陷检测是一项关键任务，其中结构可靠性和美学质量至关重要。本文提出了双线性注意力池快速区域卷积神经网络（BF-RCNN）对表面缺陷进行高精度识别和分类，以提高建筑的完整性。BF-RCNN模型的新颖性依赖于从公共库中的2400张金属板网格图像中生成神经模糊加权大津（NFO）。结合模糊逻辑和自适应阈值的思想，通过NFO滤波增强缺陷区域的可见性。对CHS图像进行BF-RCNN处理，采用ResNet-50进行深度多尺度特征提取。为了加强模型对信息特征的关注，在BF-RCNN中集成了一个挤压和激励（SE）块，实现了通道重新校准，并提高了对细微缺陷模式的灵敏度。此外，采用紧凑双线性池（CBP）对二阶特征交互进行有效建模。此外，在不增加计算成本的情况下，增强了复杂纹理和缺陷异常的表示。实验评价表明，与现有的CNN模型相比，本文提出的BF-RCNN模型的准确率、精密度和召回率分别达到99.82、99.12和98.94。提出的BF-RCNN模型为现实世界的制造和质量控制过程提供了显著的实际效益。它还可以高度可靠地识别金属板网格的表面缺陷。因此，该模型减少了人工检查时间，最大限度地减少了人为错误，并确保了一致的产品质量。该BF-RCNN为建筑行业质量检测提供了一种可扩展的智能解决方案，在缺陷评估过程中具有准确性和可靠性。

{"title":"Enhancing resilience through bilinear attention pooled region proposal network for metal sheet grid defect classification","authors":"Shyamala Devi M , Priya S , Yuvaraj Natarajan , Sri Preethaa K.R","doi":"10.1016/j.array.2025.100647","DOIUrl":"10.1016/j.array.2025.100647","url":null,"abstract":"<div><div>Defect detection in metal sheet grids is a critical task in the construction industry, where structural reliability and aesthetic quality are crucial. This paper proposes Bilinear Attention Pooled Faster Region-based Convolutional Neural Network (BF-RCNN) to identify and classify surface defects with high accuracy to enhance the building integrity. The novelty of BF-RCNN model relies on generation of Neuro fuzzy weighted otsu (NFO) from 2,400 metal sheet grid images from the public repository. By combining the idea of fuzzy logic and adaptive thresholding, the visibility of defective regions is enhanced by NFO filtering. CHS images are processed by BF-RCNN that uses ResNet-50 for deep multi-scale feature extraction. To strengthen the model's attention to informative features, a Squeeze-and-Excitation (SE) block is integrated into BF-RCNN, enabling channel-wise recalibration and improved sensitivity to subtle defect patterns. Further, Compact Bilinear Pooling (CBP) is employed to model second-order feature interactions efficiently. Also, the representation of complex textures and defect anomalies are enhanced without increasing computational cost. Experimental evaluations show that proposed BF-RCNN model achieves high accuracy, precision and recall of 99.82, 99.12 and 98.94 respectively against existing CNN models. The proposed BF-RCNN model offers significant practical benefits for real-world manufacturing and quality control processes. It also enables high reliable identification of surface defects in metal sheet grids. Thus, the model reduces manual inspection time, minimizes human error, and ensures consistent product quality. This BF-RCNN presents a scalable and intelligent solution for quality inspection in construction industry with both accuracy and reliability in defect assessment processes.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100647"},"PeriodicalIF":4.5,"publicationDate":"2025-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788318","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

Classification of adeno-associated viruses with semi-supervised learning algorithm 基于半监督学习算法的腺相关病毒分类

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-15 DOI: 10.1016/j.array.2025.100648

Al Hossain , Umme Hani Konok , Raihan Ul Islam , Mohammad Shahadat Hossain , Prashanta Dutta

Transfer learning–based deep supervised learning methods have recently become a mainstay for classification tasks. They have been applied to distinguish adeno-associated viruses (AAVs) carrying single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), or no DNA (Empty) from resistive pulse images generated by solid-state nanopore experiments. However, obtaining large quantities of labeled nanopore data, which play a crucial role in enhancing the accuracy of supervised learning, can be challenging. In this research work, we applied FixMatch—a simple yet powerful semi-supervised algorithm, which utilizes a smaller amount of labeled data to classify AAVs with ssDNA, dsDNA, or no DNA from resistive pulse sensor data. In FixMatch, the teacher model is developed by fine-tuning one of the pre-trained deep convolutional networks—GoogleNet, ResNet-50, WideResNet-50-2, and VGG-19. Using only 20% of the data as labeled and 80% as unlabeled, FixMatch achieved mean accuracies between 90% and 100% with low variance. Our results demonstrate that FixMatch can match or exceed the performance of supervised learning while dramatically reducing labeling effort. Furthermore, the prediction accuracy of FixMatch was much superior to that of MixMatch (a commonly used semi-supervised learning model). However, FixMatch's training time was notably longer compared to fully supervised approaches. Nevertheless, the high accuracy achieved with FixMatch, even with a small proportion of labeled data, opens new avenues for streamlining classification in data-scarce environments.

基于迁移学习的深度监督学习方法最近成为分类任务的主流。它们已被用于从固态纳米孔实验产生的电阻脉冲图像中区分携带单链DNA （ssDNA）、双链DNA （dsDNA）或无DNA （Empty）的腺相关病毒（aav）。然而，获得大量标记的纳米孔数据对提高监督学习的准确性起着至关重要的作用，这可能是一个挑战。在这项研究工作中，我们应用了fixmatch -一种简单但功能强大的半监督算法，该算法利用较少的标记数据从电阻脉冲传感器数据中对具有ssDNA， dsDNA或无DNA的aav进行分类。在FixMatch中，教师模型是通过微调一个预训练的深度卷积网络（googlenet, ResNet-50， WideResNet-50-2和VGG-19）来开发的。FixMatch仅使用20%的标记数据和80%的未标记数据，在低方差的情况下实现了90%到100%的平均准确率。我们的结果表明，FixMatch可以匹配或超过监督学习的性能，同时大大减少了标记工作。此外，FixMatch的预测精度远优于MixMatch（一种常用的半监督学习模型）。然而，与完全监督的方法相比，FixMatch的训练时间明显更长。尽管如此，FixMatch实现的高精度，即使是一小部分标记数据，也为在数据稀缺的环境中简化分类开辟了新的途径。

{"title":"Classification of adeno-associated viruses with semi-supervised learning algorithm","authors":"Al Hossain , Umme Hani Konok , Raihan Ul Islam , Mohammad Shahadat Hossain , Prashanta Dutta","doi":"10.1016/j.array.2025.100648","DOIUrl":"10.1016/j.array.2025.100648","url":null,"abstract":"<div><div>Transfer learning–based deep supervised learning methods have recently become a mainstay for classification tasks. They have been applied to distinguish adeno-associated viruses (AAVs) carrying single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), or no DNA (Empty) from resistive pulse images generated by solid-state nanopore experiments. However, obtaining large quantities of labeled nanopore data, which play a crucial role in enhancing the accuracy of supervised learning, can be challenging. In this research work, we applied FixMatch—a simple yet powerful semi-supervised algorithm, which utilizes a smaller amount of labeled data to classify AAVs with ssDNA, dsDNA, or no DNA from resistive pulse sensor data. In FixMatch, the teacher model is developed by fine-tuning one of the pre-trained deep convolutional networks—GoogleNet, ResNet-50, WideResNet-50-2, and VGG-19. Using only 20% of the data as labeled and 80% as unlabeled, FixMatch achieved mean accuracies between 90% and 100% with low variance. Our results demonstrate that FixMatch can match or exceed the performance of supervised learning while dramatically reducing labeling effort. Furthermore, the prediction accuracy of FixMatch was much superior to that of MixMatch (a commonly used semi-supervised learning model). However, FixMatch's training time was notably longer compared to fully supervised approaches. Nevertheless, the high accuracy achieved with FixMatch, even with a small proportion of labeled data, opens new avenues for streamlining classification in data-scarce environments.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100648"},"PeriodicalIF":4.5,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788321","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

End-to-end MLOps framework for EEG-based epilepsy prediction: Automating deep learning pipelines for scalable healthcare solutions 用于基于脑电图的癫痫预测的端到端MLOps框架：可扩展医疗保健解决方案的自动化深度学习管道

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-15 DOI: 10.1016/j.array.2025.100649

Salem Trabelsi , Salah Gontara , Khaled Ben Khalifa , Abdellatif Mtibaa

Predicting epileptic seizures from EEG signals is a critical yet challenging task, especially when transitioning machine learning (ML) models from laboratory settings to clinical environments. Most existing approaches lack adaptability to patient variability or remain confined to experimental contexts, relying on manual pipelines with limited automation. They generally do not support deployment and are seldom validated under real-world conditions, limiting scalability and clinical applicability. To address these limitations, we propose an end-to-end MLOps framework specifically designed for EEG-based seizure prediction. Our solution leverages the CRISP-ML(Q) methodology to manage the full ML lifecycle, including automated preprocessing, cross-validation, patient-specific model training, hyperparameter optimization, scalable deployment, and real-time monitoring. The predictive engine integrates a one-dimensional Convolutional Neural Network (CNN) with a Bidirectional Long Short-Term Memory (BiLSTM) network and an attention mechanism to capture spatial–temporal EEG dynamics. Validated on the CHB-MIT EEG dataset, the model achieved an average accuracy of 99.0% and sensitivity of 99.4% on training and test datasets, demonstrating strong generalization and adaptability to inter-subject variability. Following training, the MLOps pipeline autonomously selected optimal configuration parameters, including processing and model architecture, yielding performance improvements exceeding 15% compared to static baselines. In production, the deployed system remained fully operational under intensive concurrent usage, with consistent inference latency below one second in remote conditions. A user-friendly, real-time dashboard enables continuous monitoring, seizure alarm triggering, and live visualization of model behavior and evaluation metrics. These results confirm the robustness, scalability, and clinical readiness of our MLOps-driven framework for real-time EEG-based seizure prediction.

从脑电图信号预测癫痫发作是一项关键但具有挑战性的任务，特别是当机器学习（ML）模型从实验室环境过渡到临床环境时。大多数现有的方法缺乏对患者可变性的适应性，或者仍然局限于实验环境，依赖于自动化程度有限的人工管道。它们通常不支持部署，并且很少在实际条件下进行验证，从而限制了可伸缩性和临床适用性。为了解决这些限制，我们提出了一个端到端的MLOps框架，专门设计用于基于脑电图的癫痫发作预测。我们的解决方案利用CRISP-ML(Q)方法来管理整个ML生命周期，包括自动化预处理、交叉验证、特定患者模型训练、超参数优化、可扩展部署和实时监控。该预测引擎将一维卷积神经网络（CNN）与双向长短期记忆（BiLSTM）网络和注意机制相结合，捕捉脑电的时空动态。在CHB-MIT EEG数据集上验证，该模型在训练和测试数据集上的平均准确率为99.0%，灵敏度为99.4%，具有较强的泛化能力和对主体间变异性的适应性。经过训练后，MLOps管道自动选择最佳配置参数，包括处理和模型架构，与静态基线相比，性能提高超过15%。在生产中，部署的系统在密集的并发使用下保持完全运行，在远程条件下的推理延迟一致低于1秒。用户友好的实时仪表板支持连续监控、癫痫警报触发以及模型行为和评估指标的实时可视化。这些结果证实了mlops驱动的实时脑电图癫痫发作预测框架的稳健性、可扩展性和临床就绪性。

{"title":"End-to-end MLOps framework for EEG-based epilepsy prediction: Automating deep learning pipelines for scalable healthcare solutions","authors":"Salem Trabelsi , Salah Gontara , Khaled Ben Khalifa , Abdellatif Mtibaa","doi":"10.1016/j.array.2025.100649","DOIUrl":"10.1016/j.array.2025.100649","url":null,"abstract":"<div><div>Predicting epileptic seizures from EEG signals is a critical yet challenging task, especially when transitioning machine learning (ML) models from laboratory settings to clinical environments. Most existing approaches lack adaptability to patient variability or remain confined to experimental contexts, relying on manual pipelines with limited automation. They generally do not support deployment and are seldom validated under real-world conditions, limiting scalability and clinical applicability. To address these limitations, we propose an end-to-end MLOps framework specifically designed for EEG-based seizure prediction. Our solution leverages the CRISP-ML(Q) methodology to manage the full ML lifecycle, including automated preprocessing, cross-validation, patient-specific model training, hyperparameter optimization, scalable deployment, and real-time monitoring. The predictive engine integrates a one-dimensional Convolutional Neural Network (CNN) with a Bidirectional Long Short-Term Memory (BiLSTM) network and an attention mechanism to capture spatial–temporal EEG dynamics. Validated on the CHB-MIT EEG dataset, the model achieved an average accuracy of 99.0% and sensitivity of 99.4% on training and test datasets, demonstrating strong generalization and adaptability to inter-subject variability. Following training, the MLOps pipeline autonomously selected optimal configuration parameters, including processing and model architecture, yielding performance improvements exceeding 15% compared to static baselines. In production, the deployed system remained fully operational under intensive concurrent usage, with consistent inference latency below one second in remote conditions. A user-friendly, real-time dashboard enables continuous monitoring, seizure alarm triggering, and live visualization of model behavior and evaluation metrics. These results confirm the robustness, scalability, and clinical readiness of our MLOps-driven framework for real-time EEG-based seizure prediction.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100649"},"PeriodicalIF":4.5,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788316","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

EdgeSched-DQN: An intelligent deep reinforcement learning-based framework for optimized task scheduling in edge-cloud environments edgesche - dqn：一种基于深度强化学习的边缘云环境下任务调度优化框架

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-15 DOI: 10.1016/j.array.2025.100645

Nagendar Yamsani , P. Chenna Reddy

Edge-cloud computing is providing a fundamental for low-latency and resource-intensive Internet of Things (IoT) applications by creating a green ecosystem for task scheduling and resource utilization. That said, these approaches are not able to efficiently handle dynamically varying workloads, do not place resources optimally and also may not scale in an unpredictable environment. Heuristic-based methods are inflexible, conventional machine learning approach is static, while traditional reinforcement learning (RL) suffers from the issues of action space complexity and real-time deployment. To overcome these constraints, we propose a new Deep Q-Network (DQN)-based task scheduling framework, EdgeSched-DQN. An adaptive pruning principle which minimizes action space dimensionality and balances resource allocation with desired levels of system balance, and efficiency, has been integrated into the framework. An optimal task execution, which enforces latency constraints, can be achieved by a tailored reward function that accommodates changing workloads on-the-fly. Abstract It can achieve a 25 % higher reward, 20 % shorter response time, and 18 % higher success rate than state-of-the-art methods. These numbers prove the capability of EdgeSched-DQN as a competent and efficient approach for task scheduling in latency-sensitive applications. It is real-time aware when applied in IoT ecosystems, smart cities and autonomous systems to utilize limited resources more efficiently in edge-cloud environments.

边缘云计算通过为任务调度和资源利用创建绿色生态系统，为低延迟和资源密集型的物联网（IoT）应用提供了基础。也就是说，这些方法不能有效地处理动态变化的工作负载，不能最优地放置资源，也不能在不可预测的环境中扩展。基于启发式的方法缺乏灵活性，传统的机器学习方法是静态的，而传统的强化学习（RL）则存在动作空间复杂性和实时部署的问题。为了克服这些限制，我们提出了一种新的基于深度q网络（DQN）的任务调度框架，EdgeSched-DQN。一个自适应剪枝原则，使行动空间维度最小化，平衡资源分配与期望的系统平衡和效率水平，已被集成到框架中。优化的任务执行（强制延迟限制）可以通过定制的奖励功能来实现，该功能可以适应动态变化的工作负载。与现有的方法相比，该方法可以提高25%的奖励，缩短20%的响应时间，提高18%的成功率。这些数字证明了edgesche - dqn作为延迟敏感应用程序中任务调度的有效方法的能力。当应用于物联网生态系统、智慧城市和自治系统时，它可以实时感知，以便在边缘云环境中更有效地利用有限的资源。

{"title":"EdgeSched-DQN: An intelligent deep reinforcement learning-based framework for optimized task scheduling in edge-cloud environments","authors":"Nagendar Yamsani , P. Chenna Reddy","doi":"10.1016/j.array.2025.100645","DOIUrl":"10.1016/j.array.2025.100645","url":null,"abstract":"<div><div>Edge-cloud computing is providing a fundamental for low-latency and resource-intensive Internet of Things (IoT) applications by creating a green ecosystem for task scheduling and resource utilization. That said, these approaches are not able to efficiently handle dynamically varying workloads, do not place resources optimally and also may not scale in an unpredictable environment. Heuristic-based methods are inflexible, conventional machine learning approach is static, while traditional reinforcement learning (RL) suffers from the issues of action space complexity and real-time deployment. To overcome these constraints, we propose a new Deep Q-Network (DQN)-based task scheduling framework, EdgeSched-DQN. An adaptive pruning principle which minimizes action space dimensionality and balances resource allocation with desired levels of system balance, and efficiency, has been integrated into the framework. An optimal task execution, which enforces latency constraints, can be achieved by a tailored reward function that accommodates changing workloads on-the-fly. Abstract It can achieve a 25 % higher reward, 20 % shorter response time, and 18 % higher success rate than state-of-the-art methods. These numbers prove the capability of EdgeSched-DQN as a competent and efficient approach for task scheduling in latency-sensitive applications. It is real-time aware when applied in IoT ecosystems, smart cities and autonomous systems to utilize limited resources more efficiently in edge-cloud environments.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100645"},"PeriodicalIF":4.5,"publicationDate":"2025-12-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788319","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

Enhancing security in IoT networks: A multifaceted approach to vulnerability analysis and protection 增强物联网网络的安全性：漏洞分析和保护的多方面方法

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-13 DOI: 10.1016/j.array.2025.100626

Zohre Arabi , Ramin Rajabi Oskouei , Mehdi Hosseinzadeh

The rapid proliferation of the Internet of Things (IoT) has transformed modern technology by bridging the physical and digital realms. Yet, the explosive growth of connected devices—expected to surpass 50 billion by 2025—has introduced substantial security concerns. This study investigates critical vulnerabilities within IoT systems, particularly at the device and network levels, focusing on risks such as data breaches, unauthorized access, and distributed denial-of-service (DDoS) attacks. It explores the significance of implementing standardized security practices for interoperable internet-connected hardware within various environments. Despite the simplicity and feasibility of adopting such standards, many manufacturers neglect essential security protocols, leaving devices exposed. Much like pre-flight checklists in aviation, foundational security principles should be embedded into hardware design; however, innovation in this area has been largely overlooked.

We present an innovative two-phase methodology aimed at strengthening IoT security. Manufacturers often prioritize rapid deployment over protection, resulting in devices that are ill-equipped to handle sophisticated cyber threats. Conventional security approaches, based on static and generic rules, are ill-suited to the diverse, resource-constrained, and protocol-heavy IoT landscape. Our second phase involves detecting device vulnerabilities using advanced tools, such as Nmap for network probing and Binwalk for firmware analysis. Key protective measures—including secure boot processes, firmware hashing, and secure integrated circuits (ICs)—are employed to safeguard sensitive data and ensure firmware integrity. Experimental results validate the approach's effectiveness in identifying and mitigating vulnerabilities. Visual data, including port distribution charts and CVSS-based risk assessments, highlight the necessity of prioritizing high-impact threats. Although there are limitations, such as difficulties in updating legacy devices and analyzing large networks, the proposed framework significantly reduces cybersecurity risks, builds trust in IoT systems, and establishes a solid foundation for future security developments.

物联网（IoT）的快速发展通过连接物理和数字领域改变了现代技术。然而，互联设备的爆炸式增长——预计到2025年将超过500亿——引发了严重的安全问题。本研究调查了物联网系统中的关键漏洞，特别是在设备和网络层面，重点关注数据泄露、未经授权访问和分布式拒绝服务（DDoS）攻击等风险。它探讨了在各种环境中为可互操作的互联网连接硬件实现标准化安全实践的重要性。尽管采用这些标准简单可行，但许多制造商忽视了基本的安全协议，使设备暴露在外。就像航空业的飞行前检查表一样，基本的安全原则应该嵌入到硬件设计中；然而，这一领域的创新在很大程度上被忽视了。我们提出了一种创新的两阶段方法，旨在加强物联网安全。制造商往往优先考虑快速部署而不是保护，导致设备无法应对复杂的网络威胁。基于静态和通用规则的传统安全方法不适合多样化、资源受限和协议繁重的物联网环境。我们的第二阶段涉及使用高级工具检测设备漏洞，例如用于网络探测的Nmap和用于固件分析的Binwalk。关键的保护措施——包括安全引导过程、固件散列和安全集成电路（ic）——被用来保护敏感数据和确保固件完整性。实验结果验证了该方法在识别和缓解漏洞方面的有效性。可视化数据，包括港口分布图和基于cvss的风险评估，强调了优先考虑高影响威胁的必要性。尽管存在一些局限性，例如在更新旧设备和分析大型网络方面存在困难，但所提出的框架显著降低了网络安全风险，建立了对物联网系统的信任，并为未来的安全发展奠定了坚实的基础。

{"title":"Enhancing security in IoT networks: A multifaceted approach to vulnerability analysis and protection","authors":"Zohre Arabi , Ramin Rajabi Oskouei , Mehdi Hosseinzadeh","doi":"10.1016/j.array.2025.100626","DOIUrl":"10.1016/j.array.2025.100626","url":null,"abstract":"<div><div>The rapid proliferation of the Internet of Things (IoT) has transformed modern technology by bridging the physical and digital realms. Yet, the explosive growth of connected devices—expected to surpass 50 billion by 2025—has introduced substantial security concerns. This study investigates critical vulnerabilities within IoT systems, particularly at the device and network levels, focusing on risks such as data breaches, unauthorized access, and distributed denial-of-service (DDoS) attacks. It explores the significance of implementing standardized security practices for interoperable internet-connected hardware within various environments. Despite the simplicity and feasibility of adopting such standards, many manufacturers neglect essential security protocols, leaving devices exposed. Much like pre-flight checklists in aviation, foundational security principles should be embedded into hardware design; however, innovation in this area has been largely overlooked.</div><div>We present an innovative two-phase methodology aimed at strengthening IoT security. Manufacturers often prioritize rapid deployment over protection, resulting in devices that are ill-equipped to handle sophisticated cyber threats. Conventional security approaches, based on static and generic rules, are ill-suited to the diverse, resource-constrained, and protocol-heavy IoT landscape. Our second phase involves detecting device vulnerabilities using advanced tools, such as Nmap for network probing and Binwalk for firmware analysis. Key protective measures—including secure boot processes, firmware hashing, and secure integrated circuits (ICs)—are employed to safeguard sensitive data and ensure firmware integrity. Experimental results validate the approach's effectiveness in identifying and mitigating vulnerabilities. Visual data, including port distribution charts and CVSS-based risk assessments, highlight the necessity of prioritizing high-impact threats. Although there are limitations, such as difficulties in updating legacy devices and analyzing large networks, the proposed framework significantly reduces cybersecurity risks, builds trust in IoT systems, and establishes a solid foundation for future security developments.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100626"},"PeriodicalIF":4.5,"publicationDate":"2025-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921253","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

Mobile phone image-based framework for anti-copy pattern detection and classification 基于手机图像的反复制模式检测与分类框架

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-13 DOI: 10.1016/j.array.2025.100643

Joseph Smith , Zheming Zuo , Jonathan Stonehouse , Boguslaw Obara

The proliferation of high-quality printing and reproduction technologies has exacerbated product counterfeiting. Manufacturers have employed anti-copy patterns to prevent unauthorised duplication, yet their effectiveness relies on the robustness of image processing systems. This paper presents, to the best of our knowledge, the first comprehensive mobile phone image-based framework for detecting and classifying anti-copy patterns in real-world industrial scenarios. Unlike prior studies that address segmentation, quality control, or feature extraction in isolation, our contribution lies in the non-trivial integration of these modules into a validated, end-to-end system. The framework combines the Segment Anything Model with an adaptive-angle cropping mechanism for precise segmentation, incorporates no-reference image quality assessment to filter unreliable inputs, and unifies spatial and frequency-domain features for robust representation. Dimensionality reduction and clustering then manage the feature pool efficiently. Validated on a real-world dataset of over 980 annotated samples, the system achieves 99.49% classification accuracy under varied imaging conditions, demonstrating both the feasibility and industrial applicability of an integrated pipeline for combating counterfeiting.

高质量印刷和复制技术的扩散加剧了产品假冒。制造商采用反复制模式来防止未经授权的复制，但其有效性依赖于图像处理系统的鲁棒性。本文提出，据我们所知，第一个全面的基于手机图像的框架检测和分类反复制模式在现实世界的工业场景。与先前的研究不同，我们的贡献在于将这些模块集成到一个经过验证的端到端系统中，而不是孤立地处理分割、质量控制或特征提取。该框架将任意片段模型与自适应角度裁剪机制相结合以实现精确分割，结合无参考图像质量评估以过滤不可靠的输入，并统一空间和频域特征以实现鲁棒表示。然后进行降维和聚类，有效地管理特征池。在超过980个带注释样本的真实数据集上进行验证，该系统在不同成像条件下的分类准确率达到99.49%，证明了综合管道打击假冒的可行性和工业适用性。

{"title":"Mobile phone image-based framework for anti-copy pattern detection and classification","authors":"Joseph Smith , Zheming Zuo , Jonathan Stonehouse , Boguslaw Obara","doi":"10.1016/j.array.2025.100643","DOIUrl":"10.1016/j.array.2025.100643","url":null,"abstract":"<div><div>The proliferation of high-quality printing and reproduction technologies has exacerbated product counterfeiting. Manufacturers have employed anti-copy patterns to prevent unauthorised duplication, yet their effectiveness relies on the robustness of image processing systems. This paper presents, to the best of our knowledge, the first comprehensive <em>mobile phone image-based framework</em> for detecting and classifying anti-copy patterns in real-world industrial scenarios. Unlike prior studies that address segmentation, quality control, or feature extraction in isolation, our contribution lies in the <em>non-trivial integration</em> of these modules into a validated, end-to-end system. The framework combines the Segment Anything Model with an adaptive-angle cropping mechanism for precise segmentation, incorporates no-reference image quality assessment to filter unreliable inputs, and unifies spatial and frequency-domain features for robust representation. Dimensionality reduction and clustering then manage the feature pool efficiently. Validated on a real-world dataset of over 980 annotated samples, the system achieves 99.49% classification accuracy under varied imaging conditions, demonstrating both the feasibility and industrial applicability of an integrated pipeline for combating counterfeiting.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100643"},"PeriodicalIF":4.5,"publicationDate":"2025-12-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788385","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