Search Results (1,632)

Vessel detection technology based on marine remote sensing imagery is of great importance. However, it often faces challenges, such as small vessel targets, cloud occlusion, insufficient data volume, and severely imbalanced class distribution in datasets. These issues result in conventional models failing to meet the accuracy requirements for practical applications. In this paper, we first construct a novel remote sensing vessel image dataset that includes various complex scenarios and enhance the data volume and diversity through data augmentation techniques. Secondly, we address the class imbalance between foreground (small vessels) and background in remote sensing imagery from two perspectives: the sensitivity of IoU metrics to small object localization errors and the innovative design of a cost-sensitive loss function. Specifically, at the dataset level, we select vessel targets appearing in the original dataset as templates and randomly copy–paste several instances onto arbitrary positions. This enriches the diversity of target samples per image and mitigates the impact of data imbalance on the detection task. At the algorithm level, we introduce the Normalized Wasserstein Distance (NWD) to compute the similarity between bounding boxes. This enhances the importance of small target information during training and strengthens the model’s cost-sensitive learning capabilities. Ablation studies reveal that detection performance is optimal when the weight assigned to the NWD metric in the model’s loss function matches the overall proportion of small objects in the dataset. Comparative experiments show that the proposed NWD-YOLO achieves Precision, Recall, and ${\mathrm{AP}}_{50}$ scores of 0.967, 0.958, and 0.971, respectively, meeting the accuracy requirements of real-world applications. Full article

Background: Accurate diagnosis of acute appendicitis in children remains challenging due to variable presentations and limitations of existing clinical scoring systems. While machine learning (ML) offers a promising approach to enhance diagnostic precision, most prior studies have been limited by small sample sizes, single-center data, or a lack of external validation. Methods: This prospective, multicenter study included 8586 pediatric patients to develop a machine learning-based diagnostic model using routinely available clinical and hematological parameters. A separate, prospectively collected external validation cohort of 3000 patients was used to assess model performance. The Random Forest algorithm was selected based on its superior performance during model comparison. Diagnostic accuracy, sensitivity, specificity, Area Under Curve (AUC), and calibration metrics were evaluated and compared with traditional scoring systems such as Pediatric Appendicitis Score (PAS), Alvarado, and Appendicitis Inflammatory Response Score (AIRS). Results: The ML model outperformed traditional clinical scores in both development and validation cohorts. In the external validation set, the Random Forest model achieved an AUC of 0.996, accuracy of 0.992, sensitivity of 0.998, and specificity of 0.993. Feature-importance analysis identified white blood cell count, red blood cell count, and mean platelet volume as key predictors. Conclusions: This large, prospectively validated study demonstrates that a machine learning-based scoring system using commonly accessible data can significantly improve the diagnosis of pediatric appendicitis. The model offers high accuracy and clinical interpretability and has the potential to reduce diagnostic delays and unnecessary imaging. Full article

To enhance the generalization of networks and avoid redundant training efforts, cross-image hyperspectral anomaly detection (HAD) based on deep learning has been gradually studied in recent years. Cross-image HAD aims to perform anomaly detection on unknown hyperspectral images after a single training process on the network, thereby improving detection efficiency in practical applications. However, the existing approaches may require additional supervised information or stacking of networks to improve model performance, which may impose high demands on data or hardware in practical applications. In this paper, a simple and lightweight unsupervised cross-image HAD method called Variational Joint Discrimination Network (VJDNet) is proposed. We leverage the reconstruction and distribution representation ability of the variational autoencoder (VAE), learning the global and local discriminability of anomalies jointly. To integrate these representations from the VAE, a probability distribution joint discrimination (PDJD) module is proposed. Through the PDJD module, the VJDNet can directly output the anomaly score mask of pixels. To further facilitate the unsupervised paradigm, a sample pair generation module is proposed, which is able to generate anomaly samples and background representation samples tailored for the cross-image HAD task. The experimental results show that the proposed method is able to maintain the detection accuracy with only a small number of parameters. Full article

Deep learning models have obvious advantages in detecting oil spills, but the training of deep learning models heavily depends on a large number of samples of high quality. However, due to the accidental nature, unpredictability, and urgency of oil spill incidents, it is difficult to obtain a large number of labeled samples in real oil spill monitoring scenarios. Surprisingly, few-shot learning can achieve excellent classification performance with only a small number of labeled samples. In this context, a new cross-domain few-shot SAR oil spill detection network is proposed in this paper. Significantly, the network is embedded with a hybrid attention feature extraction block, which consists of a coordinate attention module to perceive the channel information and spatial location information, as well as a global self-attention transformer module capturing the global dependencies and a multi-scale self-attention module depicting the local detailed features, thereby achieving deep mining and accurate characterization of image features. In addition, to address the problem that it is difficult to distinguish between the suspected oil film in seawater and real oil film using few-shot due to the small difference in features, this paper proposes a double loss function category determination block, which consists of two parts: a well-designed category-perception loss function and a traditional cross-entropy loss function. The category-perception loss function optimizes the spatial distribution of sample features by shortening the distance between similar samples while expanding the distance between different samples. By combining the category-perception loss function with the cross-entropy loss function, the network’s performance in discriminating between real and suspected oil films is thus maximized. The experimental results effectively demonstrate that this study provides an effective solution for high-precision oil spill detection under few-shot conditions, which is conducive to the rapid identification of oil spill accidents. Full article

Traditional data-driven approaches to lithium battery state of health (SOH) estimation face the challenges of difficult feature extraction, insufficient prediction accuracy and weak generalization. To address these issues, this study proposes a novel prediction framework with transfer learning-based linear regression (LR) and a convolutional neural network (CNN) under limited data. In this framework, first, variable inertia weight-based improved particle swarm optimization for variational mode decomposition (VIW-PSO-VMD) is proposed to mitigate the volatility of the “capacity resurgence point” and extract its time-series features. Then, the T-Pearson correlation analysis is introduced to comprehensively analyze the correlations between multivariate features and lithium battery SOH data and accurately extract strongly correlated features to learn the common features of lithium batteries. On this basis, a combination model is proposed, applying LR to extract the trend features and combining them with the multivariate strongly correlated features via a CNN. Transfer learning based on temporal feature analysis is used to improve the cross-domain learning capabilities of the model. We conduct case studies on a NASA dataset and the University of Maryland dataset. The results show that the proposed method is effective in improving the lithium battery SOH estimation accuracy under limited data. Full article

Hydroxyl-terminated-polybutadiene (HTPB)-based composite solid propellants are extensively used in aerospace and defense applications due to their high energy density, thermal stability, and processability. However, the presence of highly sensitive energetic components in their formulations leads to a significant risk of accidental ignition under electrostatic discharge, posing serious safety concerns during storage, transportation, and handling. To address this issue, this study explores the prediction of electrostatic sensitivity in HTPB propellants using machine learning techniques. A dataset comprising 18 experimental formulations was employed to train and evaluate six machine learning models. Among them, the Random Forest (RF) model achieved the highest predictive accuracy (R² = 0.9681), demonstrating a strong generalization capability through leave-one-out cross-validation. Feature importance analysis using SHAP and Gini index methods revealed that aluminum, catalyst, and ammonium perchlorate were the most influential factors. These findings provide a data-driven approach for accurately predicting electrostatic sensitivity and offer valuable guidance for the rational design and safety optimization of HTPB-based propellant formulations. Full article

Flash floods are natural disasters that have significant impacts on human life and economic damage. The detection of flash floods using remote-sensing techniques provides essential data for subsequent flood-risk assessment through the preparation of flood inventory samples. In this research, a new deep-learning approach for bi-temporal flash-flood detection in Synthetic Aperture Radar (SAR) is proposed. It combines a U-Net convolutional network with a Transformer model using a compact Convolutional Tokenizer (CCT) to improve the efficiency of long-range dependency learning. The hybrid model, namely CCT-U-ViT, naturally combines the spatial feature extraction of U-Net and the global context capability of Transformer. The model significantly reduces the number of basic blocks as it uses the CCT tokenizer instead of conventional Vision Transformer tokenization, which makes it the right fit for small flood detection datasets. This model improves flood boundary delineation by involving local spatial patterns and global contextual relations. However, the method is based on Sentinel-1 SAR images and focuses on Erbil, Iraq, which experienced an extreme flash flood in December 2021. The experimental comparison results show that the proposed CCT-U-ViT outperforms multiple baseline models, such as conventional CNNs, U-Net, and Vision Transformer, obtaining an impressive overall accuracy of 91.24%. Furthermore, the model obtains better precision and recall with an F1-score of 91.21% and mIoU of 83.83%. Qualitative results demonstrate that CCT-U-ViT can effectively preserve the flood boundaries with higher precision and less salt-and-pepper noise compared with the state-of-the-art approaches. This study underscores the significance of hybrid deep-learning models in enhancing the precision of flood detection with SAR data, providing valuable insights for the advancement of real-time flood monitoring and risk management systems. Full article

Predictive formulation can help reduce the number of experiments required to reach a target cosmetic product. The performance of Large Language Models from the open source Llama family is compared with that of conventional machine learning to predict the phase stability of shampoo formulations using a recently published dataset. The predictive strength is assessed for various train dataset sizes (obtained by stratified sampling of the full dataset) and for various Large Language Model sizes (3, 8, and 70B parameters). The predictive strength is found to increase on increasing the model size, and the Large-Language-Model-based approach outperforms conventional machine learning when the train dataset is small, delivering Area Under the Receiver Operating Curve above 0.7 with as few as 20 train samples. This work illustrates the potential of Large Language Models to further reduce the number of experiments required to reach a target cosmetic formulation. Full article

Machine learning (ML) techniques are becoming increasingly valuable for modeling the transport of pollutants in plant systems. However, two challenges (small sample sizes and a lack of quantitative calculation functions) remain when using ML to predict migration in hydroponic systems. For the bioaccumulation of per- and polyfluoroalkyl substances, we studied the key factors and quantitative calculation equations based on data augmentation, ML, and symbolic regression. First, feature expansion was performed on the input data after data preprocessing; the most important step was data augmentation. The original training set was expanded nine times by combining the synthetic minority oversampling technique and a variational autoencoder. Subsequently, the four ML models were applied to the test set to predict the selected output parameters. Categorical boosting (CatBoost) had the highest prediction accuracy (R² = 0.83). The Shapley Additive Explanation values indicated that molecular weight and exposure time were the most important parameters. We applied three symbolic regression models to obtain accurate prediction equations based on the original and augmented data. Based on augmented data, the high-dimensional sparse interaction equation exhibited the highest accuracy (R² = 0.776). Our results indicate that this method could provide crucial insights into absorption and accumulation in plant roots. Full article

The rapid spread of artificial intelligence (AI) in education means that students need to master both AI literacy and personal agency. This study situates a sample of 425 Slovenian secondary technical students within a three-tier framework that maps psychological empowerment onto AI literacy outcomes within a cultural–historical activity system. The agency competence assessments yielded four profiles of student agency, ranging from fully empowered to largely disempowered. The cluster membership explained significant additional variance in AI literacy scores, supporting the additive empowerment model in an AI-rich vocational education and training context. The predictive modeling revealed that while self-efficacy, mastery-oriented motivations, and metacognitive self-regulation contributed uniquely—though small—to improving AI literacy, an unexpectedly negative relationship was identified for internal locus of control and for behavioral self-regulation focused narrowly on routines, with no significant impact observed for grit-like perseverance. These findings underscore the importance of fostering reflective, mastery-based, and self-evaluative learning dispositions over inflexible or solely routine-driven strategies in the development of AI literacy. Addressing these nuanced determinants may also be vital in narrowing AI literacy gaps observed between diverse disciplinary cohorts, as supported by recent multi-dimensional literacy frameworks and disciplinary pathway analyses. Embedding autonomy-supportive, mastery-oriented, student-centered projects and explicit metacognitive training into AI curricula could shift control inward and benefit students with low skills, helping to forge an agency-driven pathway to higher levels of AI literacy among high school students. The most striking and unexpected finding of this study is that students with a strong sense of competence—manifested as high self-efficacy—can achieve foundational AI literacy levels equivalent to those possessing broader, more holistic agentic profiles, suggesting that competence alone may be sufficient for acquiring essential AI knowledge. This challenges prevailing models that emphasize a multidimensional approach to agency and has significant implications for designing targeted interventions and curricula to rapidly build AI literacy in diverse learner populations. Full article

Recent advances in hyperspectral image (HSI) classification have demonstrated the effectiveness of hybrid architectures that integrate convolutional neural networks (CNNs) and Transformers, leveraging CNNs for local feature extraction and Transformers for global dependency modeling. However, existing fusion approaches face three critical challenges: (1) insufficient synergy between spectral and spatial feature learning due to rigid coupling mechanisms; (2) high computational complexity resulting from redundant attention calculations; and (3) limited adaptability to spectral redundancy and noise in small-sample scenarios. To address these limitations, we propose HyperTransXNet, a novel CNN-Transformer hybrid architecture that incorporates adaptive spectral-spatial fusion. Specifically, the proposed HyperTransXNet comprises three key modules: (1) a Hybrid Spatial-Spectral Module (HSSM) that captures the refined local spectral-spatial features and models global spectral correlations by combining depth-wise dynamic convolution with frequency-domain attention; (2) a Mixture-of-Experts Routing (MoE-R) module that adaptively fuses multi-scale features by dynamically selecting optimal experts via Top-K sparse weights; and (3) a Spatial-Spectral Tokens Enhancer (SSTE) module that ensures causality-preserving interactions between spectral bands and spatial contexts. Extensive experiments on the Indian Pines, Houston 2013, and WHU-Hi-LongKou datasets demonstrate the superiority of HyperTransXNet. Full article

Healthcare cost acceleration and resource allocation issues have worsened across European health systems, where a small group of patients drives excessive healthcare spending. The prediction of high-cost utilization patterns is important for the sustainable management of healthcare and focused intervention measures. The aim of our study was to derive and validate machine learning algorithms for high-cost healthcare utilization prediction based on detailed administrative data and by comparing three algorithmic methods for the best risk stratification performance. The research analyzed extensive insurance beneficiary records which compile data from health group collective funds operated by non-life insurers across EU countries, across multiple service classes. The definition of high utilization was equivalent to the upper quintile of overall health expenditure using a moderate cost threshold. The research applied three machine learning algorithms, namely logistic regression using elastic net regularization, the random forest, and support vector machines. The models used a comprehensive set of predictor variables including demographics, policy profiles, and patterns of service utilization across multiple domains of healthcare. The performance of the models was evaluated using the standard train–test methodology and rigorous cross-validation procedures. All three models demonstrated outstanding discriminative ability by achieving area under the curve values at near-perfect levels. The random forest achieved the best test performance with exceptional metrics, closely followed by logistic regression with comparable exceptional performance. Service diversity proved to be the strongest predictor across all models, while dentistry services produced an extraordinarily high odds ratio with robust confidence intervals. The group of high utilizers comprised approximately one-fifth of the sample but demonstrated significantly higher utilization across all service classes. Machine learning algorithms are capable of classifying patients eligible for the high utilization of healthcare services with nearly perfect discriminative ability. The findings justify the application of predictive analytics for proactive case management, resource planning, and focused intervention measures across private group health insurance providers in EU countries. Full article

This study proposes a deep learning-based, non-contact method for detecting elevated intraocular pressure (IOP) by integrating Scheimpflug images with corneal biomechanical features. Glaucoma, the leading cause of irreversible blindness worldwide, requires accurate IOP monitoring for early diagnosis and effective treatment. Traditional IOP measurements are often influenced by corneal biomechanical variability, leading to inaccurate readings. To address these limitations, we present a multi-modal framework incorporating CycleGAN for data augmentation, Swin Transformer for visual feature extraction, and the Kolmogorov–Arnold Network (KAN) for efficient fusion of heterogeneous data. KAN approximates complex nonlinear relationships with fewer parameters, making it effective in small-sample scenarios with intricate variable dependencies. A diverse dataset was constructed and augmented to alleviate data scarcity and class imbalance. By combining Scheimpflug imaging with clinical parameters, the model effectively integrates multi-source information to improve high IOP prediction accuracy. Experiments on a real-world private hospital dataset show that the model achieves a diagnostic accuracy of 0.91, outperforming traditional approaches. Grad-CAM visualizations identify critical anatomical regions, such as corneal thickness and anterior chamber depth, that correlate with IOP changes. These findings underscore the role of corneal structure in IOP regulation and suggest new directions for non-invasive, biomechanics-informed IOP screening. Full article

Coastal wetlands are critical ecosystems that require effective monitoring to support conservation and restoration efforts. This study evaluates the use of small unmanned aerial systems (sUAS) and multispectral imagery to estimate aboveground biomass (AGB) in tidal marshes, comparing models calibrated with destructive versus non-destructive in situ sampling methods. Imagery was collected over South Carolina’s North Inlet-Winyah Bay National Estuarine Research Reserve, and vegetation indices (VIs) were derived from sUAS imagery to model biomass. Stepwise linear regression was used to develop and validate models based on both sampling approaches. Destructive sampling models, particularly those using the Normalized Difference Vegetation Index (NDVI) and Difference Vegetation Index (DVI), achieved the lowest root mean square error (RMSE) values (as low as 70.91 g/m²), indicating higher predictive accuracy. Non-destructive models, while less accurate (minimum RMSE of 214.86 g/m²), demonstrated higher R² values (0.44 and 0.61), suggesting the potential for broader application with further refinement. These findings highlight the trade-offs between ecological impact and model performance, and support the viability of non-destructive methods for biomass estimation in sensitive wetland environments. Future work should explore machine learning approaches and improved temporal alignment of data collection to enhance model robustness. Full article

Defect detection in mass production lines often involves small and imbalanced datasets, necessitating the use of few-shot learning methods. Traditional deep learning-based approaches typically rely on large datasets, limiting their applicability in real-world scenarios. This study explores few-shot learning models for detecting product defects using limited data, enhancing model generalization and stability. Unlike previous deep learning models that require extensive datasets, our approach effectively performs defect detection with minimal data. We propose a Siamese network that integrates Residual blocks, Squeeze and Excitation blocks, and Convolution Block Attention Modules (ResNet-SE-CBAM Siamese network) for feature extraction, optimized through triplet loss for embedding learning. The ResNet-SE-CBAM Siamese network incorporates two primary features: attention mechanisms and metric learning. The recently developed attention mechanisms enhance the convolutional neural network operations and significantly improve feature extraction performance. Meanwhile, metric learning allows for the addition or removal of feature classes without the need to retrain the model, improving its applicability in industrial production lines with limited defect samples. To further improve training efficiency with imbalanced datasets, we introduce a sample selection method based on the Structural Similarity Index Measure (SSIM). Additionally, a high defect rate training strategy is utilized to reduce the False Negative Rate (FNR) and ensure no missed defect detections. At the classification stage, a K-Nearest Neighbor (KNN) classifier is employed to mitigate overfitting risks and enhance stability in few-shot conditions. The experimental results demonstrate that with a good-to-defect ratio of 20:40, the proposed system achieves a classification accuracy of 94% and an FNR of 2%. Furthermore, when the number of defective samples increases to 80, the system achieves zero false negatives (FNR = 0%). The proposed metric learning approach outperforms traditional deep learning models, such as parametric-based YOLO series models in defect detection, achieving higher accuracy and lower miss rates, highlighting its potential for high-reliability industrial deployment. Full article