Search Results (360)

In this study, we developed a leakage-free time-series machine learning framework to improve the accuracy of short-term (10 min ahead) wind speed forecasting. The measurements were obtained from real operational data collected at the Bandırma/Balıkesir wind power plant in Türkiye. The framework incorporates chronological train validation test splitting, causal missing data imputation, leakage-free feature engineering, and supervised lag-based modeling. Such a leak-proof design is crucial to avoid future information influencing the training and testing process of models, thus making the forecasting process more realistic and reliable in practice. We tested several models, including persistence, Support Vector Regression (SVR), Least-Squares Gradient Boosting (LSBoost), Random Forest (RF), Elastic Net (ELASTIC), and a stacking ensemble, and evaluated their performance using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), R-Squared (R²), bias measures, and skill scores, complemented by diagnostic analyses including residual distribution, autocorrelation, regime-based evaluation, Bland–Altman plots, and Quantile Quantile (Q-Q) plots. Our analyses showed that the Elastic Net model achieved balanced and statistically consistent performance, with a test RMSE of 0.6325 m/s, R² = 0.977, and negligible bias. Residual analysis indicated that errors were centered around zero, exhibited weak temporal dependence, and followed an approximately normal distribution in the central quantiles. Regime-based evaluation revealed that the model performed strongly in medium- and high-wind-speed conditions, while accuracy decreased under low wind speeds due to measurement uncertainty and low signal-to-noise ratios. Feature importance analysis indicated that previous wind speed was the dominant predictor, with solar irradiation and air temperature also contributing significantly. Forecast error decomposition showed that most prediction errors arose from natural atmospheric variability, with minimal systematic bias. The Diebold–Mariano test confirmed that ELASTIC statistically outperformed conventional machine learning models such as SVR and Random Forest. The proposed framework demonstrates statistically consistent short-term forecasting behavior that may support operational wind energy management and grid balancing applications. Full article

Infrared small target detection (IRSTD) is a task in target detection and computer vision that remains challenging but also critical. The cause of its complexity and difficulty lies in the inherent features of this class of targets, as most of the dataset has extreme class imbalance, weak classification contrast, and complex noise clutter in the background. Focusing on these existing issues, this work proposes CG-IRNet, a structure-aware detection framework that integrates multi-scale feature aggregation with Structure–Confidence Hybrid (SCH) loss, which integrates an augmented variant of confidence-aware Scale–Location Sensitive (SLS) loss with instance-wise structural supervision and a confidence-guided background suppression mechanism, which are all targeted towards enhancing localization consistency while largely reducing false alarms. In addition to these, a frequency-aware feature refinement module is incorporated to strengthen small target saliency under highly cluttered scenes. This work included a series of extensive experiments across three benchmark datasets included in SIRST, namely IRSTD-1K, NUAA-SIRST, and NUDT-SIRST. These experiments demonstrate a superior trade-off between detection probability (Pd) and false alarm rate. On IRSTD-1K, CG-IRNet achieves 65.09 mIoU and reduces the false alarm rate to 30.992 × 10⁻⁶, which is significantly lower than SCTransNet (55.74 × 10⁻⁶) at the same detection probability (93.27%). On NUAA-SIRST and NUDT-SIRST, the proposed method achieves 96.95% and 98.62% detection probability, respectively, while maintaining competitive or lower false alarm rates under challenging background conditions. These outcomes effectively demonstrate the improvements achieved in this work and the effectiveness of the proposed confidence-guided suppression and structure-aware optimization. Also included in the group of experiments performed in this work is the ablation study on model hyperparameters and qualitative analyses, which further confirm the joint improvements contributed by the proposed structural supervision and confidence-aware design, particularly in regimes where a low false alarm rate is the goal of optimization. Full article

Structural health monitoring of buried pipelines is essential due to their exposure to corrosion, impact loads, and geotechnical disturbances, which may induce abnormal vibration responses. Acceleration signals provide direct and sensitive measurements of buried pipeline structural dynamic behavior, and are therefore suitable for early anomaly identification. An acceleration-based intelligent framework integrating Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) and a Long Short-Term Memory (LSTM) network is proposed for buried pipeline condition recognition. First, the raw acceleration signals are decomposed into a set of intrinsic mode functions (IMFs) using ICEEMDAN to enhance time–frequency resolution and isolate weak transient impact components associated with buried pipeline structural anomalies. Subsequently, multi-scale features extracted from the IMFs are fused and fed into an LSTM network to capture temporal dependencies and perform supervised health state classification. Experimental results demonstrate that the proposed framework achieves an F1-score of 0.70 and a Precision–Recall AUC of 0.72 for identifying anomalies. Furthermore, cross-validation utilizing multi-source field data (dynamic acceleration and quasi-static strain) confirms the model’s physical interpretability and its stable performance under severe noise interference. The results validate the feasibility of combining advanced signal decomposition with deep learning techniques for buried pipeline anomaly pre-warning, providing a rigorous methodological basis for the safe operation of critical energy infrastructures. Full article

Change detection (CD) refers to the analysis of changes in the utilization of land, buildings, and other targets in the same surface environment using relevant technologies and remote sensing images. Although deep learning-based change detection methods have achieved excellent results, they remain highly dependent on extensive labeled data. High-resolution remote sensing imagery typically encompasses an abundance of details and a greater quantity of pixels compared to low-resolution datasets. Therefore, data annotation costs are significantly higher. Currently, within the context of semi-supervised change detection (SSCD) driven by consistency learning, pseudo-labels are usually selected only by threshold screening, but this ignores the spatial relationships among pixels and does not fully utilize unlabeled data, thereby affecting the model’s performance. Consequently, we propose a semi-supervised high-resolution remote sensing image change detection method based on label expansion. First, a “one weak, two strong” (OW-TS) consistency regularization (CR) framework is introduced to constrain the overall consistency between the prediction results of weak and strong augmentations, as well as between the two strong augmentations. At the same time, the location interaction map (LIM) is introduced to utilize the global–local relationship between pixels and mine the consistency of pseudo-labels, thereby improving the model’s accuracy. Empirical findings indicate that when the model is trained utilizing 20% labeled data and 80% unlabeled data on the LEVIR-CD dataset, the IoU^c index reaches 83.38%. The model performs well in smoothing the boundary between changed and unchanged areas and is comparable in performance to some fully supervised methods. Full article

Extracting roads accurately from remote sensing images is important for map updates, traffic analysis, and infrastructure monitoring. Medium-resolution multispectral images can provide useful surface and background information, but when used alone, the spatial details are limited for retaining narrow roads, intersection structures, and fine road topologies. To address this problem, this paper proposes GeoRoad-UPerNet, a Geo-1-centered weakly supervised multispectral framework for road extraction. In this framework, Geo-1 serves as the primary 16-band multispectral source, Sentinel-2 Level-2A imagery serves as auxiliary contextual support, and OpenStreetMap (OSM) road information is converted into proxy supervision rather than dense manual ground truth. GeoRoad-UPerNet contains three modules: a Geo Spectral Semantic Stem (GSSS), a Geo-Auxiliary Gated Fusion module (GAGF), and a Road Semantic Multi-Task Head (RSMH). GSSS strengthens road-sensitive multispectral responses in the Geo-1 branch. GAGF injects Sentinel-2 context through a Geo-centered gate instead of symmetric channel concatenation. RSMH imposes restrained hierarchy- and material-aware semantic regularization on the shared decoder representation during training. On the fixed source-domain benchmark, the complete model achieves an IoU of 0.7204, an F1-score of 0.8375, a Precision of 0.8092, and a Recall of 0.8678 against OSM-derived proxy masks. Relative to the UPerNet-MiT-B3 early-fusion baseline, IoU, F1-score, and Precision increase by 6.29%, 3.65%, and 12.58%, respectively. These results indicate that role-aware multisource organization improves road extraction under proxy supervision and reduces boundary noise and background false positives. Full article

Additive manufacturing (AM) monitoring is fundamentally constrained by the severe scarcity of annotated data for layer-wise segmentation. This paper addresses this bottleneck by introducing a scalable, high-fidelity synthetic data generation pipeline built on the Slice-100K dataset, capable of producing large volumes of layer-wise semantic segmentation masks. Through analysis of this large-scale synthetic data, we identify a systemic foreground–background class imbalance (1:24 ratio) inherent to AM monitoring, which causes standard Dice loss formulations to diverge catastrophically into a phenomenon we formalize as the “Dice Crash.” To effectively leverage large amounts of unlabeled data, we adapt the Meta Pseudo-Labeling (MPL) framework for industrial segmentation. We evaluate MPL’s true marginal utility by integrating it with both a standard U-Net and a robust state-of-the-art nnU-Net architecture. Experimental outputs show that while MPL yields substantial performance gains (+15.2%) on weak baselines, integrating it with an optimally configured strong baseline consistently improves segmentation accuracy and suppresses false foreground detections, thereby mitigating confirmation bias. These findings demonstrate that semi-supervised learning via continuous bilevel optimization offers a practical and robust enhancement to data-scarce additive manufacturing monitoring. Because any hidden defects in the topmost layer will be permanently buried by subsequent extrusion, this foundational layer-wise segmentation step is the most critical primitive of the monitoring pipeline. Full article

Online graduate micro-credentials are promoted both as flexible learning pathways for working professionals and as portable signals of capability for employers and professional communities. Yet, scholarship on these credentials is dispersed across policy, education, technology, and workforce literatures, making it difficult to see how the field is framed and where evidence is accumulating. This study uses OpenAlex to build an updateable evidence map of online graduate micro-credentialing. A total of 2535 records (2010–2026) were retrieved and deduplicated to 2150 works. The corpus was annotated with a transparent seedless triage step. A conservatively revised keyword typology was then applied to a typology-eligible subset, and topic modeling was used to surface candidate themes. Within the typology-eligible subset, 223 records were classifiable. Learning-first framings (66.8%) and stackable framings (58.7%) remained more common, and a 100-record hand-coded audit supported the revised rules (80.0% full-quadrant agreement). Large thematic clusters concern workforce/economic skills, engagement-oriented digital learning, and broad online teaching/learning, while smaller badge-related, infrastructure, and adjacent-domain clusters require cautious interpretation. The map points to a literature still weighted toward pathway design and implementation, but typology validation also indicates that structural framing is more mixed than the earlier always-assigned counts suggested. By making the search space and annotation logic transparent, this study provides a rerunnable baseline for cumulative qualitative synthesis and a clearer agenda for future research on how online graduate micro-credentials function as both learning experiences and credential signals. Full article

Colonoscopy polyp segmentation is important for colorectal cancer screening, yet it remains challenging because polyps exhibit large morphological variation, weak lesion–background contrast, blurred boundaries, and severe foreground–background imbalance. To address these issues, this paper presents MCM-UNet++, a hybrid U-Net++-based segmentation framework that combines three targeted enhancements. First, a Multi-Axis Transformer Block (MATransformerBlock) is incorporated into convolutional feature blocks to model long-range horizontal and vertical dependencies with lower complexity than dense global self-attention. Second, a Cross-Channel Mixing (CCM) module is used in nested skip fusion paths to recalibrate the channel and spatial responses and reduce redundant feature transmissions. Third, a Multi-Objective Adaptive Loss (MOALoss) combines focal, Dice, and boundary-aware terms with learnable weights to improve supervision for small regions and ambiguous boundaries. Experiments on four public polyp segmentation datasets (Kvasir-SEG, CVC-ClinicDB, CVC-ColonDB, and ETIS-Larib) show competitive performance against the selected baseline methods, with Dice/IoU scores of 0.9563/0.9278 on Kvasir-SEG and 0.8593/0.7896 on CVC-ColonDB. These results indicate that the proposed components can improve benchmark-level polyp segmentation performance, while broader validation is still required before clinical deployment. Full article

Accurate farmland boundary mapping from high-resolution aerial imagery is vital for precision agriculture, yet existing methods struggle with complex geospatial boundaries and texture degradation in fragmented plots. To address irreversible detail loss under downsampling, difficulty in capturing both sharp boundaries and large-scale textures, and weak boundary supervision without extra annotations, we propose PaintingFormer, an enhanced UNet-based segmentation framework. It introduces three targeted innovations: an original feature retention module (OFRM) that injects raw RGB images into the deepest decoder layer to recover lost details; a dual attention–MLP design combining FeaAttention (full-resolution global attention with linear complexity) and TWLK-MLP (cascaded 3 × 3, 5 × 5, and 7 × 7 depthwise separable kernels within an MLP) to capture multi-scale spatial patterns; and a deep edge loss from the encoder’s bottleneck that enforces boundary constraints without manual edge labels. PaintingFormer surpasses mainstream methods, achieving 84.5% mIoU and 91.5% F1 on Vaihingen, 87.3% mIoU on Potsdam, 53.7% on LoveDA, and 84.2% on our private dataset. This work offers an effective solution for fine-grained farmland segmentation, improving boundary accuracy and texture preservation. Full article

Soil salinization threatens agricultural sustainability in arid zones, yet quantitative attribution of its spatiotemporal dynamics to multi-source drivers remains scarce at regional scales. To address this, we developed an explainable framework merging Sentinel-1/2, ERA5-Land, and topographic-hydrological indices with XGBoost, trained under weak supervision with proxy labels and independently validated using field-measured ECe. A 7-group, 44-feature ensemble with spatial block 5-fold cross-validation ensured robust assessment. SHapley Additive exPlanations (SHAP) quantified driver contributions and enabled a novel dominant driver zoning (DDZ) framework. Monitoring the Hexi Corridor and Tarim Basin (2017–2024) revealed contrasting trajectories: Hexi’s dynamics were primarily climate-driven (Aridity Index), whereas 19.2% of Tarim showed significant salinization along oasis–desert margins co-dominated by elevation, soil indices, and temperature. The model achieved spatial cross-validation R² values around 0.65. DDZ mapping showed climate dominance in 98.2% of Hexi compared to 76.5% in Tarim, where terrain and optical factors were more influential. The weak supervision strategy overcomes scarce in-situ measurements, while the DDZ maps identified that Land-use-dominated zones recorded the highest salinity, offering clear directives for targeted salinity control in arid basins. Full article

Infrared small object detection (ISOD) plays a crucial role in military reconnaissance, security surveillance, and remote sensing monitoring, where weak thermal responses and complex backgrounds impose significant challenges. The recent self-supervised vision foundation model DINOv3 has demonstrated remarkable generalization ability across various visual tasks. However, directly transferring it to ISOD still remains challenging due to substantial cross-domain discrepancy between visible and infrared imagery, as well as the limited granularity of foundation features in capturing subtle thermal variations. To address these issues, this study proposes a spatial–frequency inductive bias-guided network (SFI-Net) based on DINOv3 for cross-domain representation learning in infrared small object detection. Instead of conventional domain adaptation strategies, SFI-Net explicitly models infrared-specific inductive biases in both spatial and frequency domains to enhance transferred representations. First, a spatio-frequency hybrid adapter (SFHA) is designed and embedded across multiple layers of the frozen backbone to learn infrared-specific inductive biases within distinct subspaces. Second, a feature compensation strategy with an auxiliary convolutional branch is devised to compensate for the limitation of DINOv3 in capturing multi-scale fine-grained features. Extensive experiments on the IRSTD-1K and NUDT-SIRST datasets demonstrate that the proposed SFI-Net outperforms state-of-the-art methods in both detection accuracy and computational efficiency while exhibiting strong cross-scenario generalization capability. Full article

The analysis of dynamic causal mechanisms underlying shipboard fires and explosions is often restricted by the unstructured and fragmented nature of accident investigation reports. This study proposes a framework integrating ontology-driven information extraction with Dynamic Bayesian Networks (DBNs) to model temporal accident evolution. An ontology comprising 41 nodes was constructed through a structured expert elicitation process to formalize the domain knowledge. To process 198 bilingual accident reports, an extraction pipeline was deployed, incorporating XLM-RoBERTa, BiLSTM-CRF, and an entity-marker relation classifier. Large language model (LLM)-directed weak supervision, constrained by token-level information entropy filtering, was employed to expand the training corpus, necessitating only 2.5% manual verification. The extracted semantic dependencies were utilized to initialize a three-slice DBN (precursor, initial fire, and escalation/explosion). The network structure was jointly optimized through ontology constraints (112 forbidden and 4 mandatory edges), the Hill-Climbing algorithm, and BDeu scoring. The proposed DBN achieved an AUC of 0.759 ± 0.086 and a Brier Score of 0.192 ± 0.021 (1000 bootstrap iterations), demonstrating superior predictive performance over traditional interpretable models (Static BN, HMM, ETA) with large effect sizes (Cohen’s d > 1.0), while maintaining competitive accuracy and enhanced causal interpretability relative to XGBoost. This framework offers a scalable, data-driven methodology for dynamic probabilistic risk assessment in maritime safety. Full article

The core challenge of high-precision medical image segmentation lies in modeling the multi-scale fractal self-similarity of human abdominal organs and cardiac structures, especially the blurred boundaries and weak features of low-contrast tissues. Existing CNNs and Transformers fail to simultaneously capture both global fractal topology and high-frequency fractal details, thereby limiting segmentation performance. To address this, we propose MIS-DFH, a dual-branch CNN–Transformer hybrid model that integrates Hybrid Feature Branches, Multi-Fusion Dense Frequency Skip Connections, and hierarchical Deep Supervision, achieving superior multi-scale feature extraction and segmentation performance. Experiments on the Synapse abdominal CT and ACDC cardiac MRI datasets show that MIS-DFH outperforms all compared state-of-the-art methods. Notably, it achieves 79.90% mean DSC and 20.06 mm HD95 on Synapse, representing a 5.2% DSC improvement and 34.3% HD95 reduction over MSLAU-Net, with consistent gains on ACDC. These results validate the model’s superior segmentation accuracy and clinical application value. Full article

Inverse modeling for subsurface flow represents a fundamental scientific challenge in hydrogeology and geotechnical engineering, which seeks to reconstruct critical hydrogeological parameters from sparse observational constraints. The marked spatial heterogeneity of subsurface formations, combined with the prohibitively high costs of data acquisition, renders parameter inversion, especially with very sparse supervision, inherently ill-posed and susceptible to non-uniqueness and instability. Numerical simulation-based iterative inversion methods are computationally expensive and time-consuming. Purely data-driven approaches require extensive labeled data, whereas the existing physics-informed methods lack an explicit architecture-level information transfer channel between parameter and response fields. Under sparse supervision, this prevents hydraulic head observations from effectively constraining hydraulic conductivity identification, resulting in weak parameter identifiability. In this work, we propose a physics-coupled and message-transferred inverse modeling method for transient subsurface flow problems with very sparse supervision. Specifically, the static parameter field estimated by the inversion network is explicitly incorporated into the dynamic response prediction network, and the static inversion and dynamic prediction networks are physics-coupled by the governing equations in parallel. This method enables accurate hydraulic conductivity inversion under extremely limited supervision. Experiments on multiple parameter fields, label scales, and noise levels demonstrate accurate and stable inversion performance under very sparse supervision, with ensemble-based uncertainty analysis, further confirming the reliability of the proposed method. Full article

Forward-looking sonar (FLS) is an important sensing modality for autonomous underwater vehicles and other marine robotic systems operating in turbid, low-visibility, and acoustically cluttered environments. Reliable target detection in FLS imagery remains challenging because target echoes are often weak, compact targets can be obscured by background clutter, and embedded processors impose strict limits on model size, latency, and computation. To address these issues, this study presents a lightweight FLS sensing framework for embedded target detection in resource-constrained underwater systems. The framework combines a compact detection architecture, difficulty-aware supervision, and teacher–student knowledge transfer. Specifically, FPN-Mix is developed as a lightweight backbone with a Conv-Mix module to improve contextual aggregation under limited computational budgets. A target-aware dynamic weighting loss is introduced to increase the supervision weight of difficult acoustic samples associated with weak echoes, ambiguous boundaries, and clutter interference. A multi-level knowledge distillation strategy is then adopted to transfer feature-level and prediction-level knowledge from an enhanced teacher model to the compact student detector. Experiments on the public UATD benchmark and the independently collected Zhanjiang Bay No.1 field dataset show that the proposed method achieves a favorable balance between detection accuracy and efficiency and remains competitive in a real marine aquaculture environment. The proposed model contains only 2.83 M parameters and requires 6.68 GFLOPs. After ONNX export and TensorRT FP16 acceleration, the model reaches 72.23 frames per second (FPS) on an NVIDIA Jetson Orin NX platform, supporting its practical use in embedded FLS sensing systems. Full article