Search Results (164)

Global food security faces escalating threats from climate variability and resource constraints. Accurate crop yield forecasting is essential; however, existing methods frequently overlook complex spatial dependencies driven by climate teleconnections, such as the ENSO, and lacks rigorous uncertainty quantification. This paper presents HSE-GNN-CP, a novel framework integrating heterogeneous stacked ensembles, graph neural networks (GNNs), and conformal prediction (CP). Domain-specific features are engineered, including growing degree days and climate suitability scores, and explicitly model spatial patterns via rainfall correlation graphs. The ensemble combines random forest and gradient boosting learners with bootstrap aggregation, while GNNs encode inter-regional climate dependencies. Conformalized quantile regression ensures statistically valid prediction intervals. Evaluated on a global dataset spanning 15 countries and six major crops from 1990 to 2023, the framework achieves an R² of 0.9594 and an RMSE of 4882 hg/ha. Crucially, it delivers calibrated 80% prediction intervals with 80.72% empirical coverage, significantly outperforming uncalibrated baselines at 40.03%. SHAP analysis identifies crop type and rainfall as dominant predictors, while the integrated drought classifier achieves perfect accuracy. These contributions advance agricultural AI by merging robust ensemble learning with explicit teleconnection modeling and trustworthy uncertainty quantification. Full article

Accurate forecasting of citywide electric vehicle (EV) charging load is critical for alleviating station-level congestion, improving energy dispatching, and supporting the stability of intelligent transportation systems. However, large-scale EV charging networks exhibit complex and heterogeneous spatio-temporal dependencies, and existing approaches often struggle to scale with increasing station density or long forecasting horizons. To address these challenges, we develop a modular spatio-temporal prediction framework that decouples temporal sequence modeling from spatial dependency learning under an encoder–decoder paradigm. For temporal representation, we introduce a global aggregation mechanism that compresses multi-station time-series signals into a shared latent context, enabling efficient modeling of long-range interactions while mitigating the computational burden of cross-channel correlation learning. For spatial representation, we design a dynamic multi-scale attention module that integrates graph topology with data-driven neighbor selection, allowing the model to adaptively capture both localized charging dynamics and broader regional propagation patterns. In addition, a cross-step transition bridge and a gated fusion unit are incorporated to improve stability in multi-horizon forecasting. The cross-step transition bridge maps historical information to future time steps, reducing error propagation. The gated fusion unit adaptively merges the temporal and spatial features, dynamically adjusting their contributions based on the forecast horizon, ensuring effective balance between the two and enhancing prediction accuracy across multiple time steps. Extensive experiments on a real-world dataset of 18,061 charging piles in Shenzhen demonstrate that the proposed framework achieves superior performance over state-of-the-art baselines in terms of MAE, RMSE, and MAPE. Ablation and sensitivity analyses verify the effectiveness of each module, while efficiency evaluations indicate significantly reduced computational overhead compared with existing attention-based spatio-temporal models. Full article

Research on Cooperative Driving Automation (CDA) has advanced to overcome the limited perception range of onboard sensors and the difficulty of inferring surrounding vehicles’ intentions by leveraging vehicle-to-everything (V2X) communications. This paper models how an autonomous vehicle receives cooperative sensing and cooperative maneuvering information generated at an edge roadside unit (edge RSU) that integrates roadside units (RSUs) with multi-access edge computing (MEC), and how the vehicle fuses this information with its onboard situational awareness and path-planning modules. We then analyze the performance gains of edge RSU-enabled services across diverse traffic environments. In a highway-merging scenario, simulations show that employing the edge RSU’s sensor sharing service (SSS) reduces collision risk relative to onboard-only baselines. For unsignalized intersections and roundabouts, we further propose a guidance-driven Hybrid Pairing Optimization (HPO) scheme in which the edge RSU aggregates CAV intents/trajectories, resolves spatiotemporal conflicts via lightweight pairing and time window allocation, and broadcasts maneuver guidance through MSCM. Unlike a first-come, first-served (FCFS) policy that serializes passage, HPO injects edge guidance as soft constraints while preserving arrival order fairness, enabling safe concurrent passage opportunities when feasible. Across intersections and roundabouts, HPO improves average speed by up to 192% and traffic throughput by up to 209% compared with FCFS under identical demand in our simulations. Full article

This study presents a geometry-based post-processing framework developed to refine deep-learning detections of tornado-damaged trees. The YOLO11-based instance segmentation framework served as the baseline, but its predictions often included multiple masks for a single tree or incomplete fragments of the same trunk, particularly in dense canopy areas or within tiled orthomosaics. Overlapping masks led to duplicated predictions of the same tree, while fragmentation broke a single fallen trunk into disconnected parts. Both issues reduced the accuracy of tree-count estimates and weakened orientation analysis, two factors that are critical for treefall methods. To resolve these problems, a Shape-Aware Non-Maximum Suppression (SA-NMS) procedure was introduced. The method evaluated each mask’s collinearity and, based on its geometric condition, decided whether segments should be merged, separated, or suppressed. A spatial assessment then aggregated prediction vectors within a defined Region of Interest (ROI), reconnecting trunks that were divided by obstacles or tile boundaries. The proposed method, applied to high-resolution orthomosaics from the December 2021 Land Between the Lakes tornado, achieved 76.4% and 77.1% instance-level orientation agreement accuracy in two validation zones. Full article

The inspection of power transmission lines using unmanned aerial vehicles primarily relies on object detection. However, the continuous emergence of new obstacle types necessitates frequent updates to detection models, leading to substantial retraining costs. To address this challenge, we propose a novel framework named IViT, which integrates incremental learning with a hybrid CNN-Transformer architecture for improved identification. We combined knowledge distillation with the elastic response selection distillation strategy to enhance detection performance for old classes and strengthen knowledge retention through star convolutional residual blocks constructed via element-wise multiplication. We designed a separable convolution aggregation block that integrates PConv with an attention mechanism, effectively merging global and local information to improve detection accuracy. Finally, we unified the two modules into a hybrid block. In the static detection task, IViT achieves a mAP of 55.3%, a mAP₅₀ of 83.6%, and a mAP₇₅ of 61.0%. For the incremental detection task, it attains a mAP of 57.8%, a mAP₅₀ of 79.7%, and a mAP₇₅ of 62.3%. Extensive experiments on the transmission line corridor external damage dataset and the INSPLAD dataset demonstrate that IViT exhibits outstanding detection performance compared to mainstream static object detection models and incremental object detection models. Full article

Automatic Target Recognition (ATR) in single-view sonar imagery is severely hampered by geometric distortions, acoustic shadows, and incomplete target information due to occlusions and the slant-range imaging geometry, which frequently give rise to misclassification and hinder practical underwater detection applications. To address these critical limitations, this paper proposes a Multi-View Deep Convolutional Neural Network (MVDCNN) based on feature-level fusion for robust sonar image target recognition. The MVDCNN adopts a highly modular and extensible architecture consisting of four interconnected modules: an input reshaping module that adapts multi-view images to match the input format of pre-trained backbone networks via dimension merging and channel replication; a shared-weight feature extraction module that leverages Convolutional Neural Network (CNN) or Transformer backbones (e.g., ResNet, Swin Transformer, Vision Transformer) to extract discriminative features from each view, ensuring parameter efficiency and cross-view feature consistency; a feature fusion module that aggregates complementary features (e.g., target texture and shape) across views using max-pooling to retain the most salient characteristics and suppress noisy or occluded view interference; and a lightweight classification module that maps the fused feature representations to target categories. Additionally, to mitigate the data scarcity bottleneck in sonar ATR, we design a multi-view sample augmentation method based on sonar imaging geometric principles: this method systematically combines single-view samples of the same target via the combination formula and screens valid samples within a predefined azimuth range, constructing high-quality multi-view training datasets without relying on complex generative models or massive initial labeled data. Comprehensive evaluations on the Custom Side-Scan Sonar Image Dataset (CSSID) and Nankai Sonar Image Dataset (NKSID) demonstrate the superiority of our framework over single-view baselines. Specifically, the two-view MVDCNN achieves average classification accuracies of 94.72% (CSSID) and 97.24% (NKSID), with relative improvements of 7.93% and 5.05%, respectively; the three-view MVDCNN further boosts the average accuracies to 96.60% and 98.28%. Moreover, MVDCNN substantially elevates the precision and recall of small-sample categories (e.g., Fishing net and Small propeller in NKSID), effectively alleviating the class imbalance challenge. Mechanism validation via t-Distributed Stochastic Neighbor Embedding (t-SNE) feature visualization and prediction confidence distribution analysis confirms that MVDCNN yields more separable feature representations and more confident category predictions, with stronger intra-class compactness and inter-class discrimination in the feature space. The proposed MVDCNN framework provides a robust and interpretable solution for advancing sonar ATR and offers a technical paradigm for multi-view acoustic image understanding in complex underwater environments. Full article

Rapid progress in sustainable and intelligent transportation has intensified interest in smart airport initiatives, driven by the need to support environmentally responsible and technology-enabled aviation development. As complex sociotechnical subsystems of smart aviation, smart airports integrate advanced digital, operational, and organizational technologies to enhance efficiency, resilience, and passenger experience. With increasing emphasis on such transformations, multiple strategic development plans have emerged, each with distinct priorities and implementation pathways, which necessitates a rigorous and transparent evaluation mechanism to support informed decision-making under uncertainty. This study proposes an integrated spherical fuzzy multi-criteria decision-making (MCDM) framework for assessing and ranking smart airport development plans. Subjective expert judgments are modeled using spherical fuzzy sets, allowing for the simultaneous consideration of positive, neutral, and negative membership degrees to better capture linguistic and ambiguous information. Expert importance is determined through a hybrid weighting scheme that combines a social trust network model with an entropy-based objective measure, thereby reflecting both relational credibility and informational contribution. Criterion weights are computed through an integrated approach that merges criteria importance through the inter-criteria correlation (CRITIC) method with the stepwise weight assessment ratio analysis (SWARA) method, balancing data-driven structure and expert strategic preferences. The weighted evaluations are aggregated using a spherical fuzzy extension of the combined compromise solution (CoCoSo) method to obtain the final rankings. A case study involving smart airport development planning in China is conducted to illustrate the applicability of the proposed approach. Sensitivity, ablation, and comparative analyses demonstrate that the framework yields stable, discriminative, and interpretable rankings. The results confirm that the proposed method provides a reliable and practical decision support tool for smart airport development and can be adapted to other smart transportation planning contexts. Full article

Obesity is a medical issue of increasing prevalence, with emotional eating being a key contributor to the problem, particularly in women. Theory and previous research suggest that obesity treatment participants’ self-regulatory abilities and self-efficacy to control eating are viable targets for improving emotional eating and related impacts on an unhealthy body composition. However, an improved understanding of interrelations between self-regulatory and self-efficacy changes are needed to inform behavioral treatments, which have had mostly negligible effects beyond the short term. Women were randomized into 6-month community-based obesity treatment conditions of (a) cognitive–behavioral methods with attention on emotional eating (n = 48), (b) cognitive–behavioral methods with no specific attention on emotional eating (n = 48), and (c) weight loss education (n = 50). Study-related improvements were greater in the merged cognitive–behavioral condition (n = 96; aggregated because the two corresponding treatment conditions demonstrated no significant differences). Using data aggregated across all study participants, early change in eating-related self-regulation was a significantly stronger predictor of longer-term change in eating-related self-efficacy than vice versa. Consistent with that finding, paths from treatment condition→change in self-regulation→change in self-efficacy→change in emotional eating over both 6 and 12 months were significant but not where change in self-efficacy was, instead, entered as a predictor of self-regulation change. Lessened emotional eating was significantly associated with concurrent reductions in weight and waist circumference. Consistent with self-regulation theory, findings suggest benefits for cognitive–behavioral obesity treatments over the more common education-based approaches, as well as benefits for first focusing on self-regulation that could empower increases in self-efficacy. Consistent with self-efficacy theory, such induced increases might promote favorable behavioral changes. Full article

Multi-modal object tracking (MMOT) has received widespread attention for the ability to overcome single-sensor perception limitations. However, existing methods encounter several critical challenges. Representation learning and generalization capabilities of models are constrained by the inherent heterogeneity of cross-task multi-modal data and inter-modal synergy imbalance. Particularly, in dynamically changing complex scenarios, the reliability and stability of data significantly degrade, further exacerbating the difficulty in multi-modal consistent perception and aggregation. To tackle the above issues, we propose SMUTrack, a unified framework with global shared parameters integrating three downstream MMOT tasks. SMUTrack implements a batch merging-and-splitting alternating strategy, coupled with multi-task joint training, to establish latent correlations across inter- and intra-task modalities, effectively avoiding over-reliance on certain modalities. Concurrently, we design a hierarchical modality synergy and reinforcement (HMSR) module, and a gated fusion and context awareness (GFCA) module to enable progressive multi-modal information exchange and integration, yielding the more discriminative and robust multi-modal representation. More importantly, we introduce a spatial–temporal information propagation (SIP) mechanism, which synchronously learns object trajectory cues and appearance variations to effectively build contextual relationships in long-term video tracking. Experimental results definitively validate the outstanding performance of SMUTrack on mainstream MMOT datasets, exhibiting its powerful adaptability to various MMOT tasks. Full article

Background: Cervical cancer (CC) continues to be an important public health concern in Latin America, where it is the second cause of cancer-related deaths among women. With its strong culture of vaccination, Panama was the first country to implement the HPV vaccine as part of its Essential Program on Immunization (EPI). Recently, the government implemented the 90:70:90 PAHO/WHO strategy to reach milestones toward CC elimination. Objective: This analysis triangulates and assesses national data on HPV vaccination coverage, screening practices, and cervical cancer incidence and mortality in Panama, to understand historical tendencies to date and establish a comprehensive foundation for future impact evaluations and research studies. The analysis aims to identify trends and gaps in prevention efforts and to serve as a reference point for future research on HPV-associated cancers. Methods: Population-based, descriptive, observational, ecological study where four, aggregate, de-identified data sources by various curators in Panama were match-merged by year, sex, and administrative division. Reported outcomes include HPV vaccine coverage, CC incidence and mortality rates, screening Pap tests, and CC behavior at diagnosis (in situ vs. invasive). Results: Panama has high HPV vaccine uptake (≥85% most years) in spite of low Pap test coverage (~10%). A decreasing trend in CC incidence has been observed continuously since the 1990s, counterintuitively to significantly increasing CC mortality rates, with most cases diagnosed as invasive and among younger women (30–69 years old). Conclusions: This report provides a comprehensive foundation for understanding trends in HPV vaccination coverage, cervical cancer incidence and mortality, and screening practices in Panama. While high vaccine uptake and declining incidence trends are encouraging, persistent low screening rates and elevated mortality—particularly at invasive stages among younger women—highlight critical gaps in prevention efforts. The need for integrated strategies that strengthen data systems, improve early detection, and address structural and sociocultural barriers are discussed, framed within Panama’s progress toward achieving the 90:70:90 targets. Future studies should focus on understanding non-medical influences on health and further vaccine impact with patient-level data, and other forms of HPV-related cancers in immunosuppressed populations. Public strategies would benefit from the implementation of real-life data and efficient data management, consolidation systems, systematic health promotion interventions, and an increase in resource allocation for women at the highest risk. Full article

Understanding global precipitation trends is critical for managing water resources, anticipating extreme events, and assessing the impacts of climate change. This study analyzes spatial and temporal patterns of precipitation from 1998 to 2024 using NASA’s Global Precipitation Measurement Mission (GPM) Integrated Multi-satellite Retrievals for (IMERG) Version 7, which merges satellite observations with rain-gauge data at 0.1° resolution. A total of 324 monthly datasets were aggregated into annual and seasonal composites to evaluate annual and seasonal trends in global precipitation. The non-parametric Mann–Kendall test was applied at the pixel scale to detect statistically significant monotonic trends, and Sen’s slope estimator method was used to quantify the magnitude of change in mean annual and seasonal global precipitation. Results reveal robust and geographically consistent patterns: significant wetting trends are evident in high-latitude regions, with the Arctic and Southern Oceans showing the strongest increases across multiple seasons, including +0.04 mm/day in December–January–February for the Arctic Ocean and +0.04 mm/day in June–July–August for the Southern Ocean. Northern China also demonstrates persistent increases, aligned with recent intensification of extreme late-season precipitation. In contrast, significant drying trends are detected in the tropical East Pacific (up to −0.02 mm/day), northern South America, and some areas in central-southern Africa, highlighting regions at risk of sustained hydroclimatic stress. The North Atlantic south of Greenland emerges as a summer drying hotspot, consistent with Greenland Ice Sheet melt enhancing stratification and reducing precipitation. Collectively, the findings underscore a dual pattern of wetting at high latitudes and drying in tropical belts, emphasizing the role of polar amplification, ocean–atmosphere interactions, and climate variability in shaping Earth’s precipitation dynamics. Full article

Dynamic equivalence of active distribution networks (ADNs) is emerging as one of the most important issues for the backbone network security analysis due to high penetration of distributed generations (DGs) and electricity vehicles (EVs). The multiscale and multimodal fusion deep learning (MMFDL) method proposed in this paper contains two modalities, one of which is a CNN + attention module to simulate Newton Raphson power flow calculation (NRPFC) for the important feature extraction of a power system caused by disturbance, which is motivated by the similarities between NRPFC and convolution network computation. The other is a long short-term memory (LSTM) + fully connected (FC) module for load modeling based on the fact that LSTM + FC can represent a load′s differential algebraic equations (DAEs). Moreover, to better capture the relationship between voltage and power, the multiscale fusion method is used to aggregate load modeling models with different voltage input sizes and combined with CNN + attention, merging as MMFDL to represent the dynamic behaviors of ADNs. Then, the Kepler optimization algorithm (KOA) is applied to automatically tune the adjustable parameters of MMFLD (called KOA-MMFDL), especially the LSTM and FC hidden layer number, as they are important for load modeling and there is no human knowledge to set these parameters. The performance of the proposed method was evaluated by employing different electric power systems and various disturbance scenarios. The error analysis shows that the proposed method can accurately represent the dynamic response of ADNs. In addition, comparative experiments verified that the proposed method is more robust and generalizable than other advanced non-mechanism methods. Full article

The goal of multi-view graph clustering (MVGC) for remote sensing data is to obtain a consistent partitioning by capturing complementary and consensus information across multiple views. However, numerous ambiguous background samples in multi-view remote sensing data increase structural heterogeneity while simultaneously hindering effective information extraction and fusion. Existing MVGC methods cannot selectively integrate and fully refine both graph structure and node attribute information for consensus representation learning. Furthermore, current methods tend to overlook distant nodes, thus failing to capture the global graph structure. To solve these issues, we propose a novel method called Improved Multi-View Graph Clustering with Global Graph Refinement (IMGCGGR). Specifically, we first design a view-specific fusion network (VSFN) to extract and integrate node attribute and structural information into view-specific representation for each view. VSFN not only utilizes a global self-attention mechanism to enhance the global properties of structural information but also constructs a clustering loss through a self-supervised strategy to guide the view-specific clustering distribution assignment. Moreover, to enhance the capability of view-specific representation, a learnable attention-driven aggregation strategy is introduced to flexibly fuse the attribute and structural feature. Then, we adopt a cross-view fusion module to adaptively merge multiple view-specific representations for generating the final consensus representation. Comprehensive experiments show that IMGCGGR achieves significant clustering performance improvements over baseline methods across various benchmark datasets. Full article

This paper presents a flexible approach to a multipurpose, heterogeneous archive and data management system model that merges the robustness of legacy grid-based technologies with modern cloud and edge computing paradigms. It leverages innovations driven by big data, IoT, AI, and machine learning to create an adaptive data storage and processing framework. In today’s digital age, where data are the new intangible gold, the “gold rush” lies in managing and storing massive datasets effectively—especially when these data serve governmental or commercial purposes, raising concerns about privacy and data misuse by third-party aggregators. Astronomical data, in particular, require this same thoughtful approach. Scientific discovery increasingly depends on efficient extraction and processing of large datasets. Distributed archival models, unlike centralized warehouses, offer scalability by allowing data to be accessed and processed across locations via cloud services. Incorporating edge computing further enables real-time access with reduced latency. Major astronomical projects must also avoid common single points of failure (SPOFs), often resulting from suboptimal technological choices driven by collaboration politics or In-Kind Contributions (IKCs). These missteps can hinder innovation and long-term project success. The principal goal of this work is to outline best practices in archival and data management projects—from policy development and task planning to use-case definition and implementation. Only after these steps can a coherent selection of hardware, software, or virtual environments be made. The proposed model—CTAARCHS (Cloud-based Technologies for Astronomical Archiving Research Contents and Handling Systems)—is an open-source, multidisciplinary platform supporting big data needs in astronomy. It promotes broad institutional collaboration, offering code repositories and sample data for immediate use. Full article

In recent years, detection methods for generic object detection have achieved significant progress. However, due to the large number of small objects in aerial images, mainstream detectors struggle to achieve a satisfactory detection performance. The challenges of small object detection in aerial images are primarily twofold: (1) Insufficient feature representation: The limited visual information for small objects makes it difficult for models to learn discriminative feature representations. (2) Background confusion: Abundant background information introduces more noise and interference, causing the features of small objects to easily be confused with the background. To address these issues, we propose a Multi-Level Contextual and Semantic Information Aggregation Network (MCSA-Net). MCSA-Net includes three key components: a Spatial-Aware Feature Selection Module (SAFM), a Multi-Level Joint Feature Pyramid Network (MJFPN), and an Attention-Enhanced Head (AEHead). The SAFM employs a sequence of dilated convolutions to extract multi-scale local context features and combines a spatial selection mechanism to adaptively merge these features, thereby obtaining the critical local context required for the objects, which enriches the feature representation of small objects. The MJFPN introduces multi-level connections and weighted fusion to fully leverage the spatial detail features of small objects in feature fusion and enhances the fused features further through a feature aggregation network. Finally, the AEHead is constructed by incorporating a sparse attention mechanism into the detection head. The sparse attention mechanism efficiently models long-range dependencies by computing the attention between the most relevant regions in the image while suppressing background interference, thereby enhancing the model’s ability to perceive targets and effectively improving the detection performance. Extensive experiments on four datasets, VisDrone, UAVDT, MS COCO, and DOTA, demonstrate that the proposed MCSA-Net achieves an excellent detection performance, particularly in small object detection, surpassing several state-of-the-art methods. Full article