Array最新文献

{"title":"Machine learning models to predict desktop activity recognition based on low-point gaze features","authors":"Hazem Al-Najjar ,&nbsp;Nadia Al-Rousan ,&nbsp;Hamzeh F. Assous ,&nbsp;Dania AL-Najjar","doi":"10.1016/j.array.2025.100525","DOIUrl":"10.1016/j.array.2025.100525","url":null,"abstract":"<div><div>Eye-tracking desktop activity prediction analyzes how users behave and think through their eye movements for the purpose of behavior prediction during computer use. The study examines how low-point gaze features functioning alongside machine learning (ML) models enable predictions of eight frequent desktop activities which are Debug, Browse, Play, Read, Interpret, Search, Watch and Write. The research uses simple gaze metrics obtained from 24 users through the Tobii X2-30 eye tracker for fixation count analysis along with saccade direction and gaze point statistics in order to support scalable non-intrusive deployment. The research utilized over 200,000 samples from which statistical analytics data was derived along with spatial-temporal eye movement characteristics through preprocessing. Eight well-known ML algorithms including Logistic Regression, Decision Tree and Random Forest, Neural Network and Gradient Boosting, AdaBoost, Naive Bayes and K-Nearest Neighbors received training and evaluation through 80/20 train-test split divisions. Each model conducted testing for activities through the computation of accuracy and precision and recall and F1-score metrics. Results from the evaluations show that Neural Networks coupled with Random Forests produce the best results through average performance metrics which surpass 0.91. The sustained focus activities such as Play, Read, Interpret, Watch and Write responded best to NN while RF demonstrated its strength during tasks of task-switching and problem-solving particularly in Debug and Search activities. The presented study proves the possibility of achieving accurate eye-tracking activity prediction through lightweight gaze features alongside conventional machine learning models. This investigation brings the possibility of creating real-time dynamic interfaces and user-based systems and diagnostic assessments in both consumer and clinical applications. Future research will investigate combining different models with optimization techniques to boost robustness performance in user-variable conditions and dynamic operational settings.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100525"},"PeriodicalIF":4.5,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217795","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}

{"title":"Vehicle detection on roads based on Yolov5 with multi-scale feature fusion","authors":"Longyan Xu ,&nbsp;Peilong Li ,&nbsp;Qiang Peng ,&nbsp;Yifan Zhao ,&nbsp;Lan Zhu ,&nbsp;Sanhong Yuan","doi":"10.1016/j.array.2025.100522","DOIUrl":"10.1016/j.array.2025.100522","url":null,"abstract":"<div><div>This study focuses on Yolov5-based multi-head multi-scale adaptive feature fusion for vehicle detection to enhance the intelligence and refinement of road traffic safety management. As urbanization accelerates, road traffic problems are becoming increasingly serious. Accurate vehicle detection is crucial for traffic management to detect violations, monitor traffic flow, and prevent accidents in a timely manner. This paper proposes an improved Yolov5s-G model, which enhances the detection performance for small objects and improves the retention of feature information by introducing a small-object detection layer and a Weighted Cross-scale Fusion module (WCF), and an Adaptively Spatial Feature Fusion4(ASFF4) module. These enhancements enable the model to improve detection accuracy while maintaining moderate computational complexity. Specifically, the new small-object detection layer captures positional information of small objects more effectively, while the WCF module prevents the loss of small-object information during convolution through bidirectional cross-scale link feature fusion. Additionally, the ASFF4 module utilizes adaptive spatial feature fusion to further enhance the processing capability of feature information. Experimental results demonstrate that the improved Yolov5s-G model performs well on the vehicle detection dataset, with a mAP improvement of 9.3% compared to the original Yolov5 model. Furthermore, by introducing the knowledge distillation technique, the model has been significantly enhanced in terms of lightweighting.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100522"},"PeriodicalIF":4.5,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217796","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}

{"title":"Sensor data fusion optimization in UAV integrated navigation based on matrix factorization","authors":"Ancheng Wang ,&nbsp;Yuyan Guo ,&nbsp;Hu Chen","doi":"10.1016/j.array.2025.100524","DOIUrl":"10.1016/j.array.2025.100524","url":null,"abstract":"<div><div>Although significant progress has been made in multi-source information fusion for Unmanned Aerial Vehicle integrated navigation systems, there are still obvious shortcomings in real-time redundant compression of high-dimensional heterogeneous observation data and suppression of abnormal interference in complex dynamic environments. Most existing UAV fusion methods lack robustness in complex scenarios due to either limited temporal modeling or non-adaptive regularization. To address this issue, this study proposes an integrated navigation state estimation algorithm for Unmanned Aerial Vehicle sensor data fusion optimization. The hybrid architecture integrates RPCA-based noise separation, transformer-enhanced temporal modeling, and adaptive λ regularization achieved through shallow networks. Experimental results show that when switching between multiple scenarios and the abnormal observation ratio increases to 40 % or higher, the system output error only increases by 0.044 m, significantly lower than the comparison algorithms with 0.055 m or more. In addition, the algorithm maintains an error upper bound below 0.1 m under low integrity conditions. The proposed optimization algorithm significantly improves the environmental adaptability and robustness of Unmanned Aerial Vehicle integrated navigation systems and provides theoretical and methodological support for autonomous operation of multi-sensor navigation systems in complex environments.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100524"},"PeriodicalIF":4.5,"publicationDate":"2025-09-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217807","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}

{"title":"Hierarchical graph neural network for compressed speech steganalysis","authors":"Mustapha Hemis ,&nbsp;Hamza Kheddar ,&nbsp;Mohamed Chahine Ghanem ,&nbsp;Bachir Boudraa","doi":"10.1016/j.array.2025.100510","DOIUrl":"10.1016/j.array.2025.100510","url":null,"abstract":"<div><div>Steganalysis methods based on deep learning (DL) often struggle with computational complexity and challenges in generalizing across different datasets. In the specific case of voice-over-IP (VoIP) speech streams, detection is particularly challenging because the low bit-rate encoding creates complex, relational dependencies between speech frames. Conventional DL models, which treat data as simple sequences or grids, often fail to capture these complex inter-frame dependencies effectively. To address this gap, this paper presents the first application of a graph neural network (GNN), specifically the GraphSAGE architecture, for steganalysis of compressed VoIP speech streams. The method involves straightforward graph construction from VoIP streams and employs GraphSAGE to capture hierarchical steganalysis information, including both fine-grained details and high-level patterns, thereby achieving high detection accuracy. Experimental results demonstrate that the developed approach performs well in uncovering quantization index modulation (QIM)-based steganographic patterns in VoIP signals. It achieves detection accuracy exceeding 98% even for short 0.5-second samples, and 95.17% accuracy under challenging conditions with low embedding rates, representing an improvement of 2.8% over the best-performing state-of-the-art methods. Furthermore, the model exhibits superior efficiency, with an average detection time as low as 0.016 s for 0.5-second samples—an improvement of 0.003 s. This makes it efficient for online steganalysis tasks, providing a superior balance between detection accuracy and efficiency under the constraint of short samples with low embedding rates.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100510"},"PeriodicalIF":4.5,"publicationDate":"2025-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145156176","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}

{"title":"Federated learning meets few-shot learning: A voting ensemble based combined approach to cauliflower leaf disease classification across Non-IID data distributions","authors":"MD. Zahin Muntaqim ,&nbsp;Tangin Amir Smrity ,&nbsp;Hasan Muhammad Kafi ,&nbsp;Abu Saleh Musa Miah ,&nbsp;Fahmid Al Farid ,&nbsp;Hezerul Abdul Karim ,&nbsp;Anichur Rahman","doi":"10.1016/j.array.2025.100516","DOIUrl":"10.1016/j.array.2025.100516","url":null,"abstract":"<div><div>In recent years, the classification of cauliflower <em>(Brassica oleracea var. botrytis)</em> leaf diseases has gained significant attention within agricultural research, particularly for optimizing yield and enhancing disease management. Traditional machine learning methods often rely on large, well-distributed datasets, which are difficult to obtain due to diverse data characteristics across regions and farms. To address this challenge, we proposed a novel framework that integrates Federated Learning (FL) with Few-Shot Learning (FSL) for robust cauliflower leaf disease detection, even in environments with non-identical data distributions. Our hybrid method generalizes at the task level without needing large unlabeled datasets, unlike traditional FL methods that rely heavily on data augmentation or semi-supervised techniques to make up for a lack of data. Few-Shot Learning makes it easier for each client to adapt to new disease classes with only a few samples. This makes the hybrid framework more flexible and efficient in situations where the data is not independent and identically distributed (IID). In our study, we utilized five clients, each having their own support and query sets in an n-way k-shot configuration, where each client trains on a small set of labeled data and evaluates the model on unseen data. Federated Learning facilitates collaborative, decentralized training among these clients, while Few-Shot Learning, implemented through the Reptile meta-learning algorithm, allows each client to efficiently adapt to new classes with limited samples. Moreover, we enhance prediction accuracy through ensemble voting, where predictions from multiple pre-trained deep learning models (VGG16, ResNet50V2, Xception, DenseNet169, and MobileNetV2) are combined. The ensemble voting aggregates the individual model predictions, and the most common class prediction across all models is selected, improving the overall classification performance. This ensemble mechanism is implemented after each federated round, where the model weights are aggregated and updated based on local client training, followed by a final ensemble decision for disease classification. Additionally, we have performed an ablation study on our framework to evaluate the contribution of each component (FL, FSL, and voting ensemble). Experimental results show that the ensemble model achieves an test accuracy of 95% for 2-shot, 97% for 3-shot and 4-shot, and 100% for 5-shot configurations, demonstrating the effectiveness of the framework. The results demonstrate that our Fed-FSL hybrid framework, combined with ensemble voting, provides accurate disease classification even in heterogeneous environments, offering a scalable and adaptable solution for precision agriculture and smart farming systems.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100516"},"PeriodicalIF":4.5,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145118895","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}

{"title":"Feature-driven comparison of Git and Mercurial using DBSCAN and Random Forest","authors":"Rashed Bahlool,&nbsp;Sameh Foulad,&nbsp;Sami Dagash","doi":"10.1016/j.array.2025.100519","DOIUrl":"10.1016/j.array.2025.100519","url":null,"abstract":"<div><div>Git and Mercurial are prominent distributed version control systems that differ in their underlying architecture and resource consumption. This study presents a systematic comparison between their performance using a machine learning-driven approach that measures their resource consumption based on synthetically generated repositories to systematically control key variables. To eliminate anomalies, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm was applied. The influence of repository features was determined using the Random Forest (RF) in three performance dimensions: CPU time, memory usage, and repository size. The findings indicate that Git demonstrated superior efficiency in CPU and memory usage, particularly in branching operations, while Mercurial exhibited better storage optimization and consistency, making it suitable for large-scale projects with constrained storage capacity. To interpret our results, SHapley Additive exPlanations (SHAP) was used to reveal the direction and strength of features influence that correspond to the repository characteristics. These findings offer a practical guidance for selecting version control systems based on specific project requirements.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100519"},"PeriodicalIF":4.5,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145097766","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}

{"title":"Explaining evolving outliers for uncovering key aspects of the green comparative advantage","authors":"Fabrizio Angiulli,&nbsp;Fabio Fassetti,&nbsp;Simona Nisticò,&nbsp;Luigi Palopoli","doi":"10.1016/j.array.2025.100518","DOIUrl":"10.1016/j.array.2025.100518","url":null,"abstract":"<div><div>This paper deals with the outlier explanation problem, where the goal is to find a justification explaining what makes a known outlier different from the remaining data samples. To perform this task, we propose transformation-based explanations, which are a new kind of explanations we have recently defined (Angiulli et al., 2023) that, compared to other explanation types found in the literature, provide richer insights into the explanation task at hand, allowing decision-makers to take more informed actions. However, in several interesting application cases, decision-makers are required to analyse evolving scenarios, so that the <span><math><mrow><msup><mrow><mtext>M</mtext></mrow><mrow><mn>2</mn></mrow></msup><mtext>OE</mtext></mrow></math></span> method (Angiulli et al., 2024) we had proposed cannot be directly exploited.</div><div>Therefore, in this paper, an extension of the <span><math><mrow><msup><mrow><mtext>M</mtext></mrow><mrow><mn>2</mn></mrow></msup><mtext>OE</mtext></mrow></math></span> method, called <span><math><mrow><msup><mrow><mtext>M</mtext></mrow><mrow><mn>2</mn></mrow></msup><msub><mrow><mtext>OE</mtext></mrow><mrow><mi>e</mi></mrow></msub></mrow></math></span>, is presented that constructs transformation-based explanations in evolving data scenarios. Moreover, we consider the application of the <span><math><mrow><msup><mrow><mtext>M</mtext></mrow><mrow><mn>2</mn></mrow></msup><msub><mrow><mtext>OE</mtext></mrow><mrow><mi>e</mi></mrow></msub></mrow></math></span> method to real-world environmental data. In particular, to test the effectiveness of <span><math><mrow><msup><mrow><mtext>M</mtext></mrow><mrow><mn>2</mn></mrow></msup><msub><mrow><mtext>OE</mtext></mrow><mrow><mi>e</mi></mrow></msub></mrow></math></span> in providing useful and actionable results within evolving data contexts, we gathered data concerning seven indicators related to the comparative advantage of low-carbon technologies in the 2016–2017, 2018–2019 and 2020–2021 time periods, and reshaped them in the Evolution of Green Comparative Advantage (<span><math><mtext>evGreenCA</mtext></math></span>) data collection. Then, we applied <span><math><mrow><msup><mrow><mtext>M</mtext></mrow><mrow><mn>2</mn></mrow></msup><msub><mrow><mtext>OE</mtext></mrow><mrow><mi>e</mi></mrow></msub></mrow></math></span> onto this data collection to quantitatively and qualitatively assess the interestingness of explanations it provides to decision makers in the considered application scenario.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100518"},"PeriodicalIF":4.5,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145156162","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}

{"title":"Kinetic modelling and simulation of anaerobic digestion of fibrous waste materials and slaughterhouse waste","authors":"Joseph Yankyera Kusi ,&nbsp;Ralf Müller ,&nbsp;Florian Empl ,&nbsp;Stefan Pelz ,&nbsp;Nana Sarfo Agyemang Derkyi ,&nbsp;Francis Attiogbe","doi":"10.1016/j.array.2025.100521","DOIUrl":"10.1016/j.array.2025.100521","url":null,"abstract":"<div><h3>Background</h3><div>Anaerobic digestion is a biochemical energy conversion pathway that involves a series of biochemical reactions facilitated by bacteria, leading to the breakdown of organic material into a mixture of gases in the absence of free oxygen. The necessity for modelling the kinetics of the AD of different feedstock, different pretreated feedstock, and comparing them is to establish the most appropriate model that describes a prevailing context. This study as a result, sought to comparatively model and simulate the kinetics of the AD of agricultural residue (slaughterhouse waste, fibrous waste materials and pretreated fibrous wase material mixtures).</div></div><div><h3>Method</h3><div>To ascertain the practicality of different kinetic models for different feedstock, model selection capabilities, parameter prediction strength of models, and optimisation of AD systems, four distinct kinetic models have been employed. The study also propounded hybrid models from these and applied same. It considers the same selected models for feedstock from two extremes: plant residue and animal residues, however digested with the same technology. Thus, the finding serves as a good benchmark for comparative kinetic modelling.</div></div><div><h3>Results</h3><div>Throughout both the modelling and validation phases, the estimated correlation coefficient (R²) values consistently exceeded 0.9000, signifying a robust correlation between the experimental data and the model parameters. All models exhibited competence in simulating the anaerobic digestion (AD) of the agricultural residues under investigation.</div></div><div><h3>Conclusion</h3><div>The utilization of kinetic models for the simulation and modelling of agricultural residues is thus demonstrated to be effective, as the hybrid models as well demonstrate superior performance and offer better kinetic parameter description.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100521"},"PeriodicalIF":4.5,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145109194","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}

{"title":"Unified ARP-ViT-CNN system: Hybrid deep learning approach for segmenting and classifying multiple skin cancer lesions","authors":"J.S. ThangaPurni ,&nbsp;M. Braveen","doi":"10.1016/j.array.2025.100515","DOIUrl":"10.1016/j.array.2025.100515","url":null,"abstract":"<div><div>Skin cancer continues to be one of the most difficult conditions to diagnose accurately, as different lesion types can look very similar, and the images used for diagnosis often contain noise and large variations. Many existing deep learning models, such as traditional Neural Networks and Transformer-based models, often struggle to capture both the fine-grained local details and the broader context of the image. To address these challenges, we developed a hybrid ARP-ViT-CNN model that employs a three-stream feature extraction process, where Angular Radial Partitioning (ARP) extracts geometric and structural patterns, Convolutional Neural Networks (CNN) extract subtle local features, and exploits Vision Transformers (ViT) to code long-range dependencies through self-attention mechanisms. Then, use this feature representation to develop a more advanced classification model that can differentiate several forms of skin malignancies. The Proposed model integrates a comprehensive Pre-processing pipeline, including balancing and enhancing the dataset, to address the problem of underrepresented classes and image variability. We evaluated our model on the publicly available HAM10000 dataset for multi-class skin cancer classification and segmentation. As anticipated, the ARP-ViT-CNN model beat the conventional CNN-only and Transformer-only models on the dataset, achieving an overall accuracy of 98.2% with a precision of 0.94, a recall of 0.96, and a macro-F1 score of 0.95. The ARP-ViT-CNN segmentation module for outlining lesion boundaries performed well, and was especially reliable in providing accurate boundary delineation in the presence of the complicated lesions in the test data. The overall results indicate that our ARP-ViT-CNN framework is effective, and with the ViT being suitable for global contextual learning, ARP being the mechanism by which we were making the model invariant to rotation and scale features, and this is likely contributing to the model’s tolerance to complex dermatological expressions. The ARP-ViT-CNN model has established itself as a modern approach to fully automated skin cancer imaging for automated diagnosis and a basis for future AI-powered medical imaging.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100515"},"PeriodicalIF":4.5,"publicationDate":"2025-09-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145156163","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}

{"title":"A dual-aggregation approach to fortify federated learning against poisoning attacks in IoTs","authors":"Muawya Al Dalaien ,&nbsp;Ruzat Ullah ,&nbsp;Qasem Abu Al-Haija","doi":"10.1016/j.array.2025.100520","DOIUrl":"10.1016/j.array.2025.100520","url":null,"abstract":"<div><div>Federated learning is gaining much popularity for edge devices. It offers a decentralized approach with strong privacy-preserving capabilities. It has been widely used to secure many edge devices. IoTs also utilize federated learning for an extensive range of security applications. Nevertheless, federated learning itself is also vulnerable to security threats. One such threat is poisoning attacks. Researchers have proposed many models for addressing the issue of poisoning attacks. Most of these approaches come with models based on some external technique (cryptographic or authentication technique), which adds overhead. This paper proposes a dual aggregation approach for securing federated learning. The proposed technique leverages existing machine learning techniques without introducing additional computational overhead. The approach utilizes ensemble learning, where individual client models first aggregate predictions from random forest and gradient boosting, and then the results of all the clients are further aggregated into a global model. Experimental results demonstrate that the proposed method achieves an accuracy of 91 %, highlighting its resilience against model poisoning attacks. The proposed solution provides a lightweight and efficient framework for securing IoT systems, enhancing their resilience against adversarial threats.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100520"},"PeriodicalIF":4.5,"publicationDate":"2025-09-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145156175","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}