Search Results (6,581)

High-quality synthetic text can mitigate annotation scarcity in text classification. However, standard preference optimization often produces samples that are fluent but weakly label-specific. We present Utility-weighted Direct Preference Optimization (U-DPO), a preference-optimization framework for class-conditional synthetic data generation. In U-DPO, a task-specific classifier provides a margin-based external score for each candidate generation, which is combined with an embedding-based internal similarity score to form an overall utility. These utilities are used (i) to mine preference pairs from multiple candidates per class and (ii) to weigh each DPO update by the utility gap between preferred and dispreferred samples. This design encourages the generator to concentrate on learning informative, label-discriminative preference comparisons rather than treating all pairs equally. Across two multiclass scientific-abstract benchmarks (arXiv and WOS-11967), U-DPO consistently improves downstream SciBERT classification accuracy compared with both vanilla synthetic generation and standard DPO fine-tuning, with gains up to 0.88 percentage points on arXiv and 0.83 percentage points on WOS-11967 depending on the generator. An additional GPT-4.5-based evaluation also indicates a higher mean quality score for U-DPO samples with reduced variance. Full article

Apple aroma is an important factor influencing consumers’ preferences. To understand the overall flavor characteristics of apples (Ruixue, Liangzhi, Grystal Fuji, and Guifei), volatile compounds and aroma profiles were investigated by headspace–gas chromatography–ion mobility spectrometry (HS-GC-IMS) combined with stir bar sorptive extraction (SBSE) and gas chromatography–mass spectrometry (GC-MS). The results showed that a total of 56 aroma compounds were identified by SBSE-GC-MS, and 39 aroma-active compounds were screened out using aroma intensity (AI) and odor activity value (OAV). Aroma recombination experiments showed enhanced ‘fruity’ and ‘sweet’ notes, whereas ‘floral’, ‘woody’, and ‘green’ aromas were weaker compared to the Crystal Fuji sample. Additionally, GC-IMS coupled with principal component analysis (PCA) was used to distinguish the apple samples, and partial least squares regression (PLSR) was applied to explore the correlation between sensory attributes and characteristic aroma compounds. The results indicated that Crystal Fuji exhibited the greatest correlation with the “woody” attribute, and Ruixue was highly correlated with “fruity”, “green”, and “sour” attributes, while butanoic acid, β-damascenone, butyl acetate, pentyl acetate, furfuryl alcohol, γ-decalactone, and vanillin had a significant impact on the “flower” and “sweet” attributes of Guifei. This study clarified the characteristic aroma composition of the four apple cultivars, providing data support for apple flavor quality evaluation and cultivar optimization. Full article

Accurate wind power prediction is of great significance for the dispatch, security, and stable operation of energy systems. It helps enhance the symmetry and coordination between the highly stochastic and volatile nature of the power generation supply side and the stringent requirements for stability and power quality on the grid demand side. To further enhance the accuracy of ultra-short-term wind power forecasting, this paper proposes a novel prediction framework based on multi-layer data decomposition, reconstruction, and a combined prediction model. A multi-stage decomposition and reconstruction technique is first employed to significantly reduce noise interference: the Sparrow Search Algorithm (SSA) is utilized to optimize the parameters for an initial Variational Mode Decomposition (VMD), followed by a secondary decomposition of the high-frequency components using Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN). The resulting components are then reconstructed based on Sample Entropy (SE), effectively improving the quality of the input data. Subsequently, a hybrid prediction model named IMGWO-BiTCN-BiGRU is constructed to extract spatiotemporal bidirectional features from the input sequences. Finally, simulation experiments are conducted using actual measurement data from the Sotavento wind farm in Spain. The results demonstrate that the proposed hybrid model outperforms benchmark models across all evaluation metrics, validating its effectiveness in improving forecasting accuracy and stability. Full article

The microstructure evolution law and the changes in mechanical properties of 442 ferritic stainless steel after annealing treatment at different temperatures are systematically investigated. The results show that, as the annealing temperature increases, the cold-rolled 442 ferritic stainless steel successively undergoes the process of recovery, recrystallization and grain growth, with the microstructure gradually changing from a fibrous to recrystallized structure, and the secondary phases, such as the Nb(C, N) phase, σ phase and Laves phase, precipitate. In terms of mechanical properties, the tensile strength, yield strength and Vickers hardness gradually decrease, while the elongation after fracture gradually increases. When the annealing temperature reaches 800 °C, the material exhibits the optimal comprehensive mechanical properties. The yield strength, tensile strength and elongation reach 371 MPa, 534 MPa and 31%, respectively, and the hardness is 175 HV. The fracture mode of the sample is mainly ductile fracture. EBSD analysis indicates that the strong Brass {110}<112> texture existing in the cold-rolled state gradually weakens with the annealing process, and the {111}<110>texture strengthens, thereby reducing the influence of unfavorable textures. The research results provide theoretical basis and data support for microstructure regulation and performance optimization of 442 ferritic stainless steel. Full article

Dry machining of AISI 1045 steel is attractive for sustainable manufacturing but makes it more challenging to control surface roughness R_a. This work investigates dry turning of AISI 1045 using a 2³ factorial design with three center points (11 runs) and compares a traditional power-law correlation with a quadratic response surface model (RSM). The power-law fit on log-log data explains only about 20% of the variance, whereas the quadratic RSM achieves R² ≈ 0.98 with a root-mean-square error (RMSE) of 0.62–0.77 µm based on leave-one-out cross-validation and bootstrap resampling. Feed rate S is identified as the dominant factor, while cutting speed V and depth of cut t have secondary but non-negligible interactive effects. Sobol global sensitivity indices confirm that S and S² account for more than half of the output variance. The optimized setting within the tested domain (V ≈ 83 m/min, S = 0.60 mm/rev, t = 0.10 mm) yields a predicted R_a ≈ 5.3 µm, appropriate for semi-roughing prior to grinding. The proposed framework combines small-sample RSM, Lasso regularization, uncertainty quantification and Sobol analysis to provide an uncertainty-aware model for optimizing dry-turning parameters of AISI 1045 steel. Full article

Nitrate in Baiyangdian Lake is directly linked to the sustainability of watershed ecological functions, acting as a key priority for regional ecological protection. Subsequent to the completion of a series of ecological restoration projects, its sources have undergone inevitable shifts, rendering the original pollution control framework incompatible with the new context. Thus, accurate identification of nitrate sources and their seasonal variation characteristics constitutes a core prerequisite for enhancing the targeting of pollution management. This study integrated dual stable isotopes (δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻) in water and potential source samples, along with hydrochemical data, and applied the Bayesian stable isotope mixing model (MixSIAR) to elucidate the sources of NO₃⁻ in Baiyangdian Lake. The results indicated that denitrification exerted a weak influence on the isotopic composition of NO₃⁻ in Baiyangdian Lake. Plots of the NO₃⁻/Cl⁻ versus Cl⁻ ratios for water samples and δ¹⁵N-NO₃⁻ versus δ¹⁸O-NO₃⁻ ratios for both water samples and potential sources confirmed anthropogenic sources as the primary nitrate contributors. The δ¹⁵N-NO₃⁻ vs. 1/[NO₃⁻] plot revealed that the number of NO₃⁻ sources exceeded two. The MixSIAR model demonstrated that wastewater treatment plant (WWTP) discharge was the dominant source throughout the four seasons, accounting for 49–62% with the highest contribution in winter and the lowest in summer. Soil nitrogen release contributed 19–32%, reaching its annual peak in summer. Sediment release accounted for 11–13%, maintaining a relatively low contribution across all seasons. Chemical fertilizer, manure, and sewage (M&S), and atmospheric deposition each contributed less than 6.5%, with negligible contributions. A significant reduction in the contributions of sediment release and M&S reflected the optimization effect of long-term regional ecological restoration efforts. WWTPs point source discharge and seasonal non-point source input from soil nitrogen collectively constituted the core sources of nitrate in Baiyangdian Lake. These findings provide crucial scientific support for the precise source apportionment and differentiated management of nitrate pollution in the basin. Full article

Leaf Area Index (LAI) is the total leaf area per unit of land surface area and is a crucial parameter for assessing vegetation growth and productivity. Machine learning regression algorithms are widely applied for LAI estimation. Due to spectral response variations among sensors and susceptibility of mangrove-derived variables to environmental noise suppression, obtaining sensitivity indices and optimal machine learning regression models is essential for retrieving mangrove LAI at the population scale. This study proposes a novel approach to processing and retrieving mangrove LAI data by integrating multispectral indices with machine learning methods. Box–Cox transformation and CatBoost-based feature selection were employed to obtain the optimal dataset. Random Forest (RF), Gradient Boosting Regression Trees (GBRT), and Categorical Boosting (CatBoost) algorithms were used to evaluate the accuracy of LAI retrieval from Unmanned Aerial Vehicle (UAV) and Gaofen-6 (GF-6) data. Results indicate that when LAI > 3, LAI does not immediately saturate as CVI, MTVI 2, and other indices increase, demonstrating higher sensitivity. UAV data outperformed GF-6 data in retrieving LAI for diverse mangrove populations; during model training, RF proved more suitable for small-sample datasets, while CatBoost effectively suppressed environmental noise. Both RF and CatBoost demonstrated higher robustness in estimating Avicennia marina (AM) (RF: R² = 0.704) and Aegiceras corniculatum (AC) (R² = 0.766), respectively. Spatial distribution analysis of LAI indicates that healthy AM and AC cover 85.36% and 96.67% of the area, respectively. Spartina alterniflora and aquaculture wastewater may be among the factors affecting the health of mangrove forests in the study area. LAI retrieval holds significant importance for mangrove health monitoring and risk early warning. Full article

Estimating nitrogen (N) and the corresponding crude protein (CP) content in forage crops is essential for optimizing fertilization and livestock nutrition. However, standard methods such as the Dumas and Kjeldahl techniques are destructive, costly, and impractical for field use in certain regions of developing countries. This study evaluated four non-destructive approaches—morphometric measurements, Pantone^® color scales, smartphone-based RGB analysis (ColorDetector app), and SPAD chlorophyll readings—for predicting N and CP in Megathyrsus maximus (Mombasa grass). A total of 120 samples were collected under three nitrogen fertilization levels and assessed using linear mixed-effects models with cross-validation. Morphometric variables showed poor performance (R² < 0.01), indicating low correlation with nutrient content. Pantone-based RGB models provided slightly better predictions (R² ≈ 0.30) but were limited by subjectivity and discrete data. SPAD-based models demonstrated moderate predictive accuracy (R² ≈ 0.53; RMSE ≈ 0.46%). The highest accuracy was achieved with smartphone-derived RGB data, where full RGB models reached R² = 0.60 and RMSE = 0.45%. Based on these results, a practical green color scale was developed from RGB values to support real-time, in-field nitrogen and crude protein assessment. This study highlights smartphone imaging as a scalable, low-cost, and accurate tool for non-destructive estimation of nitrogen and crude protein in tropical forages, offering an accessible alternative to laboratory methods for producers and field technicians. Full article

The efficacy of Underwater Camouflaged Object Detection (UCOD) is fundamentally constrained by severe boundary ambiguity, where biological mimicry blends targets into complex backgrounds and aquatic optical degradation erodes edge details. We propose a lightweight boundary perception detector named CAR-YOLO (Camouflage Ambiguity Resolution YOLO). Specifically, a frequency-domain dual-path mechanism (FRM-DWT/EG-IWT) leverages selective wavelet aggregation and dynamic injection to recover high-frequency edges. Subsequently, these high-frequency cues are synergized with low-frequency semantic information via the Low-level Adaptive Fusion (LAF) module. To further address noisy samples, an Uncertainty Calibration Head (UCH) refines supervision via prediction consistency. Finally, we constructed specialized datasets based on public data for training and evaluation, including UCOD10K and UWB-COT220. On UCOD10K, CAR-YOLO achieves 27.1% mAP50–95, surpassing several state-of-the-art (SOTA) methods while reducing parameters from 2.58 M to 2.43 M and GFLOPs from 6.3 to 5.9. On the challenging UWB-COT220 benchmark, the model attains 30.7% mAP50–95, marking a 7.7-point improvement over YOLOv11. Furthermore, cross-domain experiments on UODD demonstrate strong generalization. These results indicate that CAR-YOLO effectively mitigates boundary ambiguity, achieving an optimal balance between accuracy, robustness, and efficiency. Full article

Background/Objectives: Acute leukemia, the most common childhood cancer, poses a significant public health challenge in low- and middle-income countries (LMICs) due to its high incidence and mortality rates. Survival rates in these regions are often lower, primarily due to delayed and inaccurate diagnoses, limited access to treatment, therapy abandonment, therapy-related toxicity, and inadequate healthcare infrastructure. In Mexico, a new initiative called OncoCREAN has been developed to address this urgent need by establishing local treatment centers near pediatric patients’ home cities, ensuring timely cancer detection and comprehensive disease treatment. Methods: A retrospective observational study was conducted on pediatric patients treated at the Mexican Social Security Institute (IMSS) between 18 May 2022 and 30 June 2025. Patients presenting clinical suspicion of acute leukemia were referred to OncoCREAN centers for sample collection and subsequent shipment to the Oncoimmunology and Cytomics Laboratory (OCL), where immunophenotyping confirmed the diagnoses. Results: The implementation of the OncoCREAN model significantly reduced diagnostic turnaround times, facilitating timely therapeutic decisions, minimized uncertainty, and optimized clinical management. The decentralized framework demonstrated feasibility across diverse geographic regions, ensuring access to advanced diagnostic technology for vulnerable populations and generating valuable data on disease incidence and molecular profiles. Conclusions: The OncoCREAN model highlights the critical importance of decentralizing high-technology diagnostic resources in modern pediatric oncology. This new approach to translational research that is accessible, inclusive, and relevant to society creates a paradigm shift in the management of childhood cancer and other diseases. Full article

To address the challenge of defect detection in tempered glass panel production rising from sample scarcity, this paper proposes a few-shot detection methodology that integrates an enhanced Stable Diffusion model with Mask R-CNN. Specifically, the approach utilizes a Mask Encoder to optimize the Stable Diffusion architecture, employing the Structural Similarity Index Measure (SSIM) to evaluate sample quality. This process generates high-fidelity virtual samples to construct a hybrid dataset for training data augmentation. Furthermore, a resource isolation strategy is adopted to facilitate online detection using an improved semi-supervised Mask R-CNN framework. Experimental results demonstrate that the proposed scheme effectively resolves detection difficulties for eight defect types, including edge chipping and scratches. The method achieves an mAP50 of 81.5%, representing a nearly 47% improvement over baseline methods relying solely on real samples, thereby realizing high-precision and high-efficiency industrial defect detection. Full article

In order to meet the urgent needs of refined geological disaster risk assessment at a county scale, and in view of the shortcomings of existing methods in the aspects of sample dependence, rainfall time-varying differences, and vulnerability quantification, this study takes Linxiang City as an example, integrates multi-source data such as geology, geography, meteorology, remote sensing, and field survey, and explores practical methods. A random forest (RF) model was implemented for geological hazard susceptibility mapping, and its hyper-parameters were tuned using Bayesian optimization. Based on a statistical analysis of the frequency of historical disaster events, a risk classification of rainfall in the flood season and non-flood season was evaluated. A vulnerability simplification method based on the value and exposure of disaster-bearing bodies was proposed. Finally, rapid risk assessment was achieved by matrix superposition. The results showed that the model had high accuracy (AUC = 0.903). The use of field survey risk types effectively enhanced the susceptibility sample set and verified the accuracy of risk assessment. The risk factor in the flood season and non-flood season was significantly different, and the very-high- and high-risk areas in the flood season were mainly distributed in the shallow metamorphic rock mountainous area in the east of Yanglousi Town and the granite residual soil area in the south of Zhanqiao Town, the latter of which was highly consistent with the field survey results. This study demonstrated value in terms of sample enhancement, model optimization, consideration of time-varying rainfall, and vulnerability simplification. The evaluation results can provide direct support for the construction of a “point–area dual control” system for geological disasters in Linxiang City, and the methodological framework can also provide a practical reference for risk evaluation in other counties. Full article

Background: Osteoporosis is a major complication in older adults with type 2 diabetes mellitus (T2DM). Malnutrition contributes to bone loss, and the Geriatric Nutritional Risk Index (GNRI) has emerged as a simple tool for assessing nutritional status. Evidence on the predictive value of the GNRI for osteoporosis in elderly patients with T2DM remains limited. Objective: To evaluate the association between GNRI scores and osteoporosis and determine its predictive performance in elderly patients with T2DM. Methods: A cross-sectional study was conducted on 200 elderly patients with T2DM attending the internal medicine outpatient clinics at Zagazig university hospitals between January and October 2025. Clinical data, biochemical parameters, and bone mineral density (BMD) at the lumbar spine, femoral neck, and total hip were assessed. GNRI scores were calculated using standard formulas. Participants were classified into osteoporosis and non-osteoporosis groups according to WHO criteria. Correlations and ROC curve analysis were performed to assess the predictive ability of the GNRI in comparison with age, BMI, and serum albumin. Results: Osteoporosis was present in 15% of the cohort. Patients with osteoporosis had significantly lower GNRI scores and lower BMD values at all measured sites (p < 0.05). The GNRI showed significant positive correlations with BMD parameters in both sexes. ROC analysis demonstrated that the GNRI had the highest predictive performance for osteoporosis (AUC = 0.80 for all patients; AUC = 0.85 in males; AUC = 0.77 in females). Optimal GNRI cutoff values were <100.03 for the total sample, <99.10 for males, and <100.3 for females. Conclusions: The GNRI is a valuable and simple clinical tool for predicting osteoporosis in elderly patients with T2DM. Lower GNRI scores are significantly associated with reduced BMD. Incorporating the GNRI into routine assessment may help identify high-risk patients who require early screening and intervention. Full article

In survey sampling, the presence of non-response and measurement error often leads to biased and inefficient estimates, particularly in stratified random sampling designs. This study introduces a new calibration estimation technique for stratified sampling that effectively accounts for non-response and measurement error. By incorporating auxiliary data and optimizing calibrated weights, the proposed estimator minimizes bias and enhances efficiency. The estimator employs auxiliary information through calibrated weights derived using a chi-square-type distance function. Furthermore, the performance of the suggested calibration estimator has been compared with that of the Hansen and Hurwitz’ estimator, the separate ratio-type estimator and the Singh’s estimator. To validate the efficiency and superiority of the proposed method over traditional estimators, an empirical evaluation has been carried out using simulated datasets. The comparative assessment with existing estimators demonstrates that the proposed method provides improved precision and robustness. Full article

Urban air pollution remains a critical public health issue, especially in densely populated cities where pedestrians experience direct exposure to traffic-related and environmental emissions. This study develops and tests a pollution-aware pedestrian routing framework for Thessaloniki, Greece, designed to minimize environmental exposure while maintaining route efficiency. The framework combines high-resolution air-quality data and computational techniques to represent pollution patterns at pedestrian scale. Air-quality is expressed as a continuous European Air Quality Index (EAQI) and is embedded in a network-based routing engine (OSRM) that balances exposure and distance through a weighted optimization function. Using 3000 randomly sampled origin-destination pairs, exposure-aware routes are compared with conventional shortest-distance paths across short, medium, and long walking trips. Results show that exposure-aware routes reduce cumulative AQI exposure by an average of 4% with only 3% distance increase, while maintaining stable scaling across all route classes. Exposure benefits exceeding 5% are observed for approximately 8% of medium-length routes and 24% of long routes, while short routes present minimal or no detours, but lower exposure benefits. These findings confirm that integrating high-resolution environmental data into pedestrian navigation systems is both feasible and operationally effective, providing a practical foundation for future real-time, pollution-aware mobility services in smart cities. Full article