Search Results (13,035)

Modulation recognition, as one of the key technologies in the field of wireless communications, holds significant importance in applications such as spectrum resource management, interference suppression, and cognitive radio. While deep learning has substantially improved the performance of Automatic Modulation Recognition (AMR), it heavily relies on large amounts of labeled data. Given the high annotation costs and privacy concerns, researching semi-supervised AMR methods that leverage readily available unlabeled data for training is of great significance. This study constructs a semi-supervised AMR method based on dual-student. Specifically, we first adopt a dual-branch co-training architecture to fully exploit unlabeled data and effectively learn deep feature representations. Then, we develop a dynamic stability evaluation module using strong and weak augmentation strategies to improve the accuracy of generated pseudo-labels. Finally, based on the dual-student semi-supervised framework and pseudo-label stability evaluation, we propose a stability-guided consistency regularization constraint method and conduct semi-supervised AMR model training. The experimental results demonstrate that the proposed DualBranch-AMR method significantly outperforms traditional supervised baseline approaches on benchmark datasets. With only 5% labeled data, it achieves a recognition accuracy of 55.84%, reaching over 90% of the performance of fully supervised training. This validates the superiority of the proposed method under semi-supervised conditions. Full article

This study analyses the physical performance variables involved in different training tasks aimed at replicating the worst-case scenarios (WCSs) observed during official matches in professional football, with a focus on playing positions and occurrences within a 1 min period. Data were collected from 188 training sessions and 42 matches of a Spanish Second Division team during the 2021/2022 season. All data were reported on a per-player basis. GPS tracking devices were used to record physical variables such as total distance, high-speed running (HSR), sprints, accelerations, decelerations, and high metabolic load distance (HMLD). Players were grouped according to their match positions: central defenders, wide players, midfielders and forwards. The results showed that none of the training tasks fully replicated the physical demands of match play. However, task TYPEs 11 (Large-Sided Games) and 9 (small-sided games with orientation and transition) were the closest to match demands, particularly in terms of accelerations and decelerations. Although differences were observed across all variables, the most pronounced discrepancies were observed in sprint and HSR variables, where training tasksfailed to reach 60% of match demands. These findings highlight the need to design more specific drills that simulate the intensity of WCS, allowing for more accurate weekly training load planning. This study offers valuable contributions for optimising performance and reducing injury risk in professional footballers during the competitive period. Full article

Background/Objectives: The study aims to identify key cognitive and non-cognitive variables (e.g., clinical, neuroimaging, and genetic data) predicting cognitive decline in Parkinson’s disease (PD) patients using machine learning applied to a sample (N = 618) from the Parkinson’s Progression Markers Initiative database. Traditional research has mainly employed explanatory approaches to explore variable relationships, rather than maximizing predictive accuracy for future cognitive decline. In the present study, we implemented a predictive framework that integrates a broad range of baseline cognitive, clinical, genetic, and imaging data to accurately forecast changes in cognitive functioning in PD patients. Methods: An artificial neural network was trained on baseline data to predict general cognitive status three years later. Model performance was evaluated using 5-fold stratified cross-validation. We investigated model interpretability using explainable artificial intelligence techniques, including Shapley Additive Explanations (SHAP) values, Group-Wise Feature Masking, and Brute-Force Combinatorial Masking, to identify the most influential predictors of cognitive decline. Results: The model achieved a recall of 0.91 for identifying patients who developed cognitive decline, with an overall classification accuracy of 0.79. All applied explainability techniques consistently highlighted baseline MoCA scores, memory performance, the motor examination score (MDS-UPDRS Part III), and anxiety as the most predictive features. Conclusions: From a clinical perspective, the findings can support the early detection of PD patients who are more prone to developing cognitive decline, thereby helping to prevent cognitive impairments by designing specific treatments. This can improve the quality of life for patients and caregivers, supporting patient autonomy. Full article

Machine learning (ML)-based techniques have received significant attention in various fields of industry and science. In civil and bridge engineering, they can facilitate the identification of specific patterns through the analysis of data acquired from structural health monitoring (SHM) systems. To evaluate the prediction capabilities of ML, this study examines the performance of several ML algorithms in estimating the total weight and location of vehicles on a bridge using strain sensing. A novel framework based on a combined model and data-driven approach is described, consisting of the establishment of the finite element (FE) model, its updating according to load testing results, and data augmentation to facilitate the training of selected physics-informed regression models. The article discusses the design of the Fiber Bragg Grating (FBG) sensor-based Bridge Weigh-in-Motion (BWIM) system, specifically focusing on several supervised regression models of different architectures. The current work proposes the use of the updated FE model to generate training data and evaluate the accuracy of regression models with the possible exclusion of selected input features enabled by the structural specificity of a bridge. The data were sourced from the SHM system installed on a network arch bridge in Wolin, Poland. It confirmed the possibility of establishing the BWIM system based on strain measurements, characterized by a reduced number of sensors and a satisfactory level of accuracy in the estimation of loads, achieved by exploiting the network arch bridge structural specificity. Full article

Street networks provide an invaluable source of information about the different temporal and spatial patterns emerging in our cities. These streets are often represented as graphs where intersections are modeled as nodes and streets as edges between them. Previous work has shown that raster representations of the original data can be created through a learning algorithm on low-dimensional representations of the street networks. In contrast, models that capture high-level urban network metrics can be trained through convolutional neural networks. However, the detailed topological data is lost through the rasterization of the street network, and the models cannot recover this information from the image alone, failing to capture complex street network features. This paper proposes a model capable of inferring good representations directly from the street network. Specifically, we use a variational autoencoder with graph convolutional layers and a decoder that generates a probabilistic, fully connected graph to learn latent representations that encode both local network structure and the spatial distribution of nodes. We train the model on thousands of street network segments and use the learned representations to generate synthetic street configurations. Finally, we proposed a possible application to classify the urban morphology of different network segments, investigating their common characteristics in the learned space. Full article

Background and Objectives: Peripheral vascular access in infants is a frequent but technically challenging procedure due to the anatomical characteristics of this population. Repeated failed attempts may increase complications and emotional stress for both patients and healthcare professionals. This systematic review aimed to evaluate the efficacy and safety of ultrasound-guided peripheral vascular cannulation compared to the conventional or “blind” technique in infants. Materials and Methods: A systematic review was conducted in accordance with PRISMA guidelines. The PubMed database was searched for studies published between 2017 and 2025. Studies comparing both techniques in infants under two years of age were selected, evaluating variables such as the number of punctures, first-attempt success, healthcare staff perception, associated stress, and the role of simulation in training. Results: Eleven studies were included, comprising clinical trials, observational studies, and training program assessments from different countries. Most reported a higher first-attempt success rate with the ultrasound-guided technique (often exceeding 85%), along with fewer punctures and complications, particularly among less-experienced professionals. Improvements in staff perception were also observed following structured training. The impact on stress experienced by patients and families was less frequently assessed directly, although some studies reported indirect benefits. Conclusions: Ultrasound-guided peripheral vascular cannulation appears to be more effective and safer than the conventional technique in infants, particularly in complex or critical care contexts. Its implementation requires specific training and appropriate resources but could significantly improve clinical outcomes and the pediatric patient experience. Full article

Background/Objectives: The major question that confronts a pathologist when evaluating a lymph node biopsy is whether the process is benign or malignant, and the differential diagnosis between follicular lymphoma and reactive lymphoid tissue can be challenging. Methods: This study designed a convolutional neural network based on ResNet architecture to classify a large series of 221 cases, including 177 follicular lymphoma and 44 reactive lymphoid tissue/lymphoid hyperplasia, which were stained with hematoxylin and eosin (H&E). Explainable artificial intelligence (XAI) methods were used for interpretability. Results: The series included 1,004,509 follicular lymphoma and 490,506 reactive lymphoid tissue image-patches at 224 × 244 × 3, and was partitioned into training (70%), validation (10%), and testing (20%) sets. The performance of the training (training and validation sets) had an accuracy of 99.81%. In the testing set, the performance metrics achieved an accuracy of 99.80% at the image-patch level for follicular lymphoma. The other performance parameters were precision (99.8%), recall (99.8%), false positive rate (0.35%), specificity (99.7%), and F1 score (99.9%). Interpretability was analyzed using three methods: grad-CAM, image LIME, and occlusion sensitivity. Additionally, hybrid partitioning was performed to avoid information leakage using a patient-level independent validation set that confirmed high classification performance. Conclusions: Narrow artificial intelligence (AI) can perform differential diagnosis between follicular lymphoma and reactive lymphoma tissue, but it is task-specific and operates within limited constraints. The trained ResNet convolutional neural network (CNN) may be used as transfer learning for larger series of cases and lymphoma diagnoses in the future. Full article

Youth Mental Health First Aid (YMHFA) is a widely adopted professional development tool that helps educators across the United States improve their mental health literacy. Data from a pretest/posttest evaluation of the YMHFA training delivered at five schools are used to explore whether various YMHFA outcomes differ for teachers who have and have not received previous mental health training. Specifically, the current study compares scores on confidence, knowledge, negative attitudes, and intentions to intervene prior to completing the YMHFA program (i.e., at baseline), the rate of change in each measure, and satisfaction with the training across teachers with and without previous mental health prevention training. Our findings showed that teachers with previous training scored higher on confidence, mental health knowledge, and intentions to intervene at baseline and experienced different patterns of change after completion of the YMHFA training program, compared to teachers without prior training. Negative attitudes and training satisfaction did not reveal differences across training experiences. Study findings offer important program and policy implications about teachers’ training experiences, as well as the value of implementing YMHFA as a universal training in educational settings. Full article

The automated detection and classification of immature macauba (Acrocomia aculeata) fruits is critical for improving post-harvest processing and quality control. In this study, we present a comparative evaluation of two state-of-the-art YOLO architectures, YOLOv11x and YOLOv12x, trained on the newly constructed VIC01 dataset comprising 1600 annotated images captured under both background-free and natural background conditions. Both models were implemented in PyTorch and trained until the convergence of box regression, classification, and distribution-focal losses. Under an IoU (intersection over union) threshold of 0.50, YOLOv11x and YOLOv12x achieved an identical mean average precision (mAP₅₀) of 0.995 with perfect precision and recall or TPR (true positive rate). Averaged over IoU thresholds from 0.50 to 0.95, YOLOv11x demonstrated superior spatial localization performance (mAP_50–95 = 0.973), while YOLOv12x exhibited robust performance in complex background scenarios, achieving a competitive mAP_50–95. Inference throughput averaged 3.9 ms per image for YOLOv11x and 6.7 ms for YOLOv12x, highlighting a trade-off between speed and architectural complexity. Fused model representations revealed optimized layer fusion and reduced computational overhead (GFLOPs), facilitating efficient deployment. Confusion-matrix analyses confirmed YOLOv11x’s ability to reject background clutter more effectively than YOLOv12x, whereas precision–recall and F1-score curves indicated both models maintain near-perfect detection balance across thresholds. The public release of the VIC01 dataset and trained weights ensures reproducibility and supports future research. Our results underscore the importance of selecting architectures based on application-specific requirements, balancing detection accuracy, background discrimination, and computational constraints. Future work will extend this framework to additional maturation stages, sensor fusion modalities, and lightweight edge-deployment variants. By facilitating precise immature fruit identification, this work contributes to sustainable production and value addition in macauba processing. Full article

As a high-market-value sector, the consumer electronics industry is particularly vulnerable to reputational damage from surface defects in shipped products. However, the high level of automation and the short product life cycles in this industry make defect sample collection both difficult and inefficient. This challenge has led to a severe shortage of publicly available, comprehensive datasets dedicated to surface defect detection, limiting the development of targeted methodologies in the academic community. Most existing datasets focus on general-purpose object categories, such as those in the COCO and PASCAL VOC datasets, or on industrial surfaces, such as those in the MvTec AD and ZJU-Leaper datasets. However, these datasets differ significantly in structure, defect types, and imaging conditions from those specific to consumer electronics. As a result, models trained on them often perform poorly when applied to surface defect detection tasks in this domain. To address this issue, the present study introduces a specialized optical sampling system with six distinct lighting configurations, each designed to highlight different surface defect types. These lighting conditions were calibrated by experienced optical engineers to maximize defect visibility and detectability. Using this system, 14,478 high-resolution defect images were collected from actual production environments. These images cover more than six defect types, such as scratches, plain particles, edge particles, dirt, collisions, and unknown defects. After data acquisition, senior quality control inspectors and manufacturing engineers established standardized annotation criteria based on real-world industrial acceptance standards. Annotations were then applied using bounding boxes for object detection and pixelwise masks for semantic segmentation. In addition to the dataset construction scheme, commonly used semantic segmentation methods were benchmarked using the provided mask annotations. The resulting dataset has been made publicly available to support the research community in developing, testing, and refining advanced surface defect detection algorithms under realistic conditions. To the best of our knowledge, this is the first comprehensive, multiclass, multi-defect dataset for surface defect detection in the consumer electronics domain that provides pixel-level ground-truth annotations and is explicitly designed for real-world applications. Full article

Background/Objectives: The accurate identification of insect bites from images of skin is daunting due to the fine gradations among diverse bite types, variability in human skin response, and inconsistencies in image quality. Methods: For this work, we introduce DeepBiteNet, a new ensemble-based deep learning model designed to perform robust multiclass classification of insect bites from RGB images. Our model aggregates three semantically diverse convolutional neural networks—DenseNet121, EfficientNet-B0, and MobileNetV3-Small—using a stacked meta-classifier designed to aggregate their predicted outcomes into an integrated, discriminatively strong output. Our technique balances heterogeneous feature representation with suppression of individual model biases. Our model was trained and evaluated on a hand-collected set of 1932 labeled images representing eight classes, consisting of common bites such as mosquito, flea, and tick bites, and unaffected skin. Our domain-specific augmentation pipeline imputed practical variability in lighting, occlusion, and skin tone, thereby boosting generalizability. Results: Our model, DeepBiteNet, achieved a training accuracy of 89.7%, validation accuracy of 85.1%, and test accuracy of 84.6%, and surpassed fifteen benchmark CNN architectures on all key indicators, viz., precision (0.880), recall (0.870), and F1-score (0.875). Our model, optimized for mobile deployment with quantization and TensorFlow Lite, enables rapid on-client computation and eliminates reliance on cloud-based processing. Conclusions: Our work shows how ensemble learning, when carefully designed and combined with realistic data augmentation, can boost the reliability and usability of automatic insect bite diagnosis. Our model, DeepBiteNet, forms a promising foundation for future integration with mobile health (mHealth) solutions and may complement early diagnosis and triage in dermatologically underserved regions. Full article

The integration of unmanned aerial vehicle (UAV) remote sensing and deep learning has emerged as a highly effective strategy for inventorying forest resources. However, the spatiotemporal variability of forest environments and the scarcity of annotated data hinder the performance of conventional supervised deep-learning models. To overcome these challenges, this study has developed efficient tree (ET), a semi-supervised tree detector designed for forest scenes. ET employed an enhanced YOLO model (YOLO-Tree) as a base detector and incorporated a teacher–student semi-supervised learning (SSL) framework based on pseudo-labeling, effectively leveraging abundant unlabeled data to bolster model robustness. The results revealed that SSL significantly improved outcomes in scenarios with sparse labeled data, specifically when the annotation proportion was below 50%. Additionally, employing overlapping cropping as a data augmentation strategy mitigated instability during semi-supervised training under conditions of limited sample size. Notably, introducing unlabeled data from external sites enhances the accuracy and cross-site generalization of models trained on diverse datasets, achieving impressive results with F1, mAP50, and mAP50-95 scores of 0.979, 0.992, and 0.871, respectively. In conclusion, this study highlights the potential of combining UAV-based RGB imagery with SSL to advance tree species identification in heterogeneous forests. Full article

Introduction: Strength training, essential for health and performance, often uses free weights for greater stabilization demands and pulleys for easier load adjustment and progression. Methods: The aim of the study was to analyze the differences in force application using gravitational and pneumatic resistances. Twenty experienced subjects participated in the study (age: 21.9 ± 3.8 years; body mass: 76.3 ± 9.4 kg; height: 177.4 ± 7.5 cm), performing four exercises with each type of resistance: bench press, lat pulldown, chest fly, and single-arm row. The participants performed 8 repetitions per exercise. Peak and mean force were measured with a 100 Hz load cell (SUIFF S2 Pro) during the concentric phase of the lifts. Differences between resistance types were analyzed using one-way ANOVA and paired t-tests. Results: Peak force was higher with gravitational resistance across all exercises (p < 0.001; d = 2.1–4.7). Average force with gravitational resistance was also higher in the bench press and lat pulldown (p < 0.05; d = 0.7–1.4), but not in the chest fly or single-arm row. Conclusions: Gravitational resistance may better enhance peak strength, while pneumatic resistance supports consistent force and neuromuscular control. These results allow us to select the resistance type based on specific mechanical characteristics. Full article

The increasing complexity of cybersecurity risks highlights the critical need for novel teaching techniques that provide students with the necessary skills and information. Traditional on-premises laboratory setups frequently lack the scalability, flexibility, and accessibility necessary for efficient training in today’s dynamic world. This study compares the efficacy of cloud-based solutions—specifically, Amazon Web Services (AWS) Elastic Compute Cloud (EC2)—against traditional settings like VirtualBox, with the goal of determining their potential to improve cybersecurity education. The study conducts systematic experimentation to compare lab environments based on parameters such as lab completion time, CPU and RAM use, and ease of access. The results show that AWS EC2 outperforms VirtualBox by shortening lab completion times, optimizing resource usage, and providing more remote accessibility. Additionally, the cloud-based strategy provides scalable, cost-effective implementation via a pay-per-use model, serving a wide range of pedagogical needs. These findings show that incorporating cloud technology into cybersecurity curricula can lead to more efficient, adaptable, and inclusive learning experiences, thereby boosting pedagogical methods in the field. Full article

Electricity markets in Europe have undergone major changes in the last decade, mainly due to the increasing share of variable renewable energy sources (RES), changing demand patterns, and geopolitical factors—particularly the war in Ukraine, tensions over energy imports, and disruptions in natural gas supplies. These changes have led to increased electricity price volatility, reducing the reliability of traditional forecasting tools. This research analyses the potential of static and adaptive linear regression as electricity price forecasting tools in the context of three countries with different energy mixes: Germany, France and the Czech Republic. The static regression approach was compared with an adaptive approach based on incremental model updates at monthly intervals. Testing was carried out in three different scenarios combining stable and turbulent market periods. The quantitative results showed that the adaptive model achieved a lower MAE and RMSE, especially when trained on data from high-volatility periods. However, models trained under turbulent conditions performed poorly in stable environments due to a shift in market dynamics. The results supported several of the hypotheses formulated and demonstrated the need for localised, flexible and continuously updated forecasting. Limitations of the adaptive approach and suggestions for future research, including changing the length of training windows and the use of seasonal models, are also discussed. The research confirms that modern markets require adaptive analytical approaches that account for changing RES dynamics and country specificities. Full article