Search Results (2,223)

Robust vehicle detection in real-world traffic surveillance remains challenging due to degraded imagery caused by motion blur, adverse weather, and low illumination, which significantly increases detector sensitivity to hyperparameter configurations. This study proposes a “Frugal AI” distributed multi-GPU framework that optimizes hyperparameters via a stochastic simplex-based search coupled with five-fold cross-validation. Utilizing three low-cost NVIDIA GTX 1050 Ti GPUs, the framework performs parallel candidate exploration with an asynchronous model-level exchange mechanism to escape local optima without the overhead of gradient synchronization. Seven CNN backbones—VGG16, VGG19, GoogLeNet, MobileNetV2, ResNet18, ResNet50, and ResNet101—were evaluated within YOLOv2 and Faster R-CNN detectors. To address memory constraints (4 GB VRAM), YOLOv2 was selected for extensive benchmarking. Performance was measured using a harmonic precision–recall-based cost metric to strictly penalize imbalanced outcomes. Experimental results demonstrate that under identical wall-clock time budgets, the proposed framework achieves an average 1.38% reduction in aggregated cost across all models, with the highly sensitive VGG19 backbone showing a 4.00% improvement. Benchmarking against Bayesian optimization, genetic algorithms, and random search confirms that our method achieves superior optimization quality with statistical significance ($p < 0.05$). Under a rigorous IoU = 0.75 threshold, the optimized models consistently yielded F1-scores 0.8444 $\pm$ 0.0346. Ablation studies further validate that the collaborative model exchange is essential for accelerating convergence in rugged loss landscapes. This research offers a practical, scalable, and cost-efficient solution for deploying robust AI surveillance in resource-constrained smart city infrastructure. Full article

Bike sharing is an established component of urban mobility infrastructure, offering a low-emission alternative to motorized transport for short trips in cities worldwide. Accurate demand forecasting is essential for efficient system operation: it enables better bike redistribution, reduces user wait times, and lowers the operational costs associated with rebalancing. This study evaluated multiple ensemble strategies for hourly bike-sharing demand prediction, comparing bagging methods (Random Forest, Extra Trees), boosting methods (AdaBoost, Gradient Boosting Regressor, Histogram-based Gradient Boosting Regressor), and a Voting ensemble, while systematically investigating the impact of hyperparameter optimization. A repeated hold-out protocol was used, in which the dataset was randomly divided into 80% training and 20% test subsets across 10 random splits; 5-fold cross-validation was applied within each training fold exclusively for hyperparameter tuning, ensuring the test set remained unseen during model selection. Random Search and Bayesian Optimization were compared under identical budgets of 60 configurations per model. Results show that optimization substantially improves all models, with the most pronounced gains for AdaBoost (58% RMSE reduction) and Gradient Boosting Regressor (45% RMSE reduction). A Voting ensemble combining a Random Search-tuned Gradient Boosting Regressor and a Bayesian-optimized Histogram-based Gradient Boosting Regressor achieves the best overall performance (RMSE of 38.48, R² of 0.955) with the lowest variance among all repeated splits. Feature importance analysis confirms that hour of day and temperature are the dominant demand drivers, consistent with the operational patterns of urban bike-sharing systems. The performance difference between Random Search and Bayesian Optimization is negligible for most models, suggesting that well-designed search spaces allow simpler strategies to achieve competitive results. A controlled comparison conducted under identical experimental conditions shows that the Voting ensemble is statistically equivalent to XGBoost and nominally better than LightGBM, while CatBoost achieves a statistically significant advantage, highlighting it as a strong individual alternative. Full article

Synthetic datasets are increasingly used in education research for methodological validation, privacy-preserving data sharing, and reproducible equity analysis; however, most generative approaches prioritize marginal distributional similarity without ensuring preservation of multilevel inferential properties. This limitation is consequential for repeated-measures data analyzed using intersectionality-focused hierarchical models, where conclusions depend on variance partitioning, partial pooling, and stratum-level heterogeneity. We introduce MAI-GAN, a hybrid generative framework that implements a structure–residual decomposition approach combining Bayesian longitudinal MAIHDA with conditional GAN-based residual generation. Inferential fidelity is operationalized with respect to multilevel intersectional models by explicitly targeting the preservation of fixed effects, variance components, and variance partitioning coefficients, while baseline composition is maintained via stratified bootstrap resampling. Applied to a six-semester undergraduate biology dataset (N = 2669 students), MAI-GAN was evaluated across multiple independent random seeds and consistently reproduced baseline-dependent residual structure and key inferential quantities. These results demonstrate that model-aligned generative strategies can produce synthetic longitudinal datasets that remain coherent under intersectionality-focused multilevel analysis, offering a principled foundation for equity-oriented synthetic data generation. Full article

Background: Metabolic syndrome is associated with early impairments in cellular bioenergetics that are not fully captured by conventional body composition measures. Molecular hydrogen, produced endogenously through gut microbial fermentation and measurable in breath, has been implicated in redox and mitochondrial regulation. Whether breath hydrogen relates to preservation of intracellular, metabolically active tissue in metabolic syndrome remains unclear. Objectives: To examine the association between breath hydrogen concentration and an integrated cellular bioenergetic phenotype derived from intracellular body composition indices in sedentary adults with metabolic syndrome. Methods: Twenty-eight sedentary, middle-aged adults (51.2 ± 7.9 years, 19 females) with metabolic syndrome underwent fasting breath hydrogen assessment and multifrequency bioelectrical impedance analysis. A composite cellular bioenergetic phenotype was derived using principal component analysis of body cell mass, intracellular water, total body potassium, and glycogen. Associations between breath hydrogen and the composite phenotype were evaluated using Spearman correlation with bootstrapped confidence intervals, Theil-Sen regression, and Bayesian linear regression adjusted for age, sex, and waist circumference. Sensitivity analyses included fat-free mass. Results: A single principal component explained 98.6% of the variance across intracellular variables, indicating a highly coherent cellular bioenergetic phenotype. Breath hydrogen concentration was positively associated with this phenotype (ρ = 0.43, p = 0.021; BCa 95% CI 0.07–0.70). Theil-Sen regression confirmed a robust positive association (β = 0.017 per ppm hydrogen; 95% CI 0.002–0.046). Bayesian models showed posterior distributions centered on positive effect sizes, independent of central adiposity. In contrast, the association with fat-free mass alone was borderline. Conclusions: Breath hydrogen concentration reflects an integrated intracellular bioenergetic phenotype in sedentary adults with metabolic syndrome, tracking cellular quality rather than lean mass quantity. Breath hydrogen may serve as a non-invasive biomarker of cellular bioenergetic integrity and a potential tool for phenotype-guided metabolic interventions. Full article

(1) Background: Potentially arthropod-pathogenic and plant-associated Metarhizium fungi are of high interest for basic research, biological pest control and plant growth promotion. Unambiguous species delineation enabling the taxonomic assignment of new isolates and the identification of new Metarhizium species is of crucial importance for both research and application. Recently, the new species Metarhizium hybridum and Metarhizium parapingshaense were introduced on the basis of phylogenomic studies. (2) Methods: Neighbor- joining and Bayesian inference-based phylogenetic reconstruction of ribosomal intergenic spacer (rIGS) sequences were used to critically evaluate new species introductions. A species-discriminating diagnostic PCR tool for Metarhizium was adapted to M. hybridum and M. parapingshaense. GenBank database mining was performed to identify cryptic descriptions of the new species. (3) Results: The introduction of M. hybridum and M. parapingshaense was corroborated by rIGS sequence comparison. Data mining revealed cryptic first descriptions of M. hybridum from Canada, China, Colombia, Costa Rica, Cuba, Honduras, Mexico, New Zealand, the USA and the Philippines, and of M. parapingshaense from China, India, Japan, the Philippines and South Korea. (4) Conclusions: Results support the reliability of rIGS as a single taxonomic marker for species-level identification of Metarhizium fungi. Species-discriminating diagnostic PCR was successfully adapted to enable the sequencing-independent identification of the confirmed new species M. hybridum and M. parapingshaense. Full article

Environmental and financial datasets often display complex distributional characteristics, including heavy tails, high skewness and the presence of extreme observations. Traditional probability models such as the exponential, gamma or log-normal distributions may not adequately capture these behaviours particularly when modelling extreme events such as rainfall, pollution levels, stock returns or loss severities. By integrating the characteristics of Pareto and exponential distributions into an exponentiated framework that can describe datasets arising from environmental and finance fields, this study presents a novel three-parameter exponentiated Pareto exponential distributions using the exponentiated Pareto family of distributions with classical exponential distribution as the baseline model. This novel model extends the classical exponential distribution with the addition of extra shape parameters which simultaneously regulate the centre and tail behaviours of the new model. The statistical and mathematical characteristics of the proposed distribution are determined and studied. The maximum likelihood estimate approach is used in a conducted simulation exercise, and the estimator’s efficiency is evaluated as seen from the results. The practical applicability of the model is illustrated with four real-life datasets utilising model adequacy and goodness-of-fit measurements such as log–likelihood, Akaike information criteria and Bayesian information criteria. The data reveal that the proposed model gives a better fit than the models chosen as comparators, making the EPE distribution useful and robust in environmental and financial fields of study. Full article

Accurate prediction of a ship’s Estimated Time of Arrival (ETA) is of great significance for port scheduling, logistics management, and navigation safety. Traditional ETA prediction approaches often rely on manual experience for parameter tuning, which tends to be inefficient and susceptible to subjective factors. To address this issue and improve prediction accuracy, this study proposes a hybrid modeling framework, integrating Bayesian Optimization (BO), Convolutional Neural Networks (CNNs), and Long Short-Term Memory (LSTM) networks. In this approach, Automatic Identification System (AIS) data is leveraged to predict the total voyage duration before departure, thereby deriving the vessel’s ETA. The model, referred to as BO-CNN-LSTM, utilizes BO for automatic hyperparameter tuning, employs CNN for extracting local features, and applies LSTM network to capture temporal dependencies. The model is developed using a dataset of 32,972 distinct voyage records, among which 23,947 are retained as valid samples after data cleaning. Pearson correlation analysis is conducted to select key input variables, including navigation speed, ship type, sailing distance, and deadweight tonnage. Additionally, sailing distance is processed using the Ramer–Douglas–Peucker algorithm. Experimental evaluation indicates that the BO-CNN-LSTM model achieves a coefficient of determination of 0.987, along with a mean absolute error and root mean square error of 6.078 and 8.730, respectively. These results significantly outperform comparison models such as CNN, LSTM, CNN-LSTM, random forest, AdaBoost, and Elman neural networks. Overall, this study validates the effectiveness and superiority of the proposed BO-CNN-LSTM model in ship ETA prediction, providing an efficient and effective prediction solution for intelligent maritime transportation systems. Full article

This study quantifies the spatiotemporal evolution of non-communicable disease (NCD) mortality risk associated with temperature extremes in the Metropolitan Area of the Valley of Mexico (MAVM) from 2000 to 2019. Using a Bayesian risk assessment framework, we analyzed 747,131 deaths to evaluate the impact of extreme temperature indices (Tn90p, Tn10p, TNn, Tx90p, Tx10p, TXx, DTR) across demographic and geographic dimensions. Results reveal a significant intensification of mortality risk, particularly for circulatory and metabolic diseases after 2005 and 2014. Risk expansion analysis identified 16 cases of robust relative risk (RR) intensification, predominantly among elderly populations. Females and males aged 65+ with metabolic diseases exhibited the highest thermal vulnerability. Our analysis further indicates a systematic shift in mortality risk toward higher nocturnal temperatures and reduced diurnal variability, suggesting a transition from cold-related stress to persistent nighttime heat exposure. Spatial Bayesian modeling shows a progressive homogenization of environmental risk across the metropolitan area, with high-risk thermal profiles expanding from the urban core toward peripheral municipalities, reducing the extent of previously lower-risk zones. Notably, the number of municipalities in the highest risk category for females aged 65+ with metabolic diseases increased by 550%, while for males of the same age, the expansion reached 163%. These findings indicate that vulnerability in megacities is a dynamic process driven by nocturnal warming and thermal instability. They highlight the urgent need to integrate climate-sensitive planning strategies—such as the identification and preservation of climatic refuge zones—into urban development policies, alongside continuous monitoring of temperature-related health risks. Full article

Robust measurement of entrepreneurial competencies (ECs) is crucial for entrepreneurship education, yet their internal structure remains theoretically contested and empirically underexamined. This study examined whether the four-factor Entrepreneurial Career Competency Instrument (ECCI) exhibits a hierarchical (bifactor) structure among South African entrepreneurs. Using two non-probability samples (N = 1305; N = 280), we analysed competing models, including a bifactor exploratory structural equation model (ESEM). The selected 56-item bifactor ESEM solution was examined for conceptual replicability in the smaller sample using Bayesian structural equation modelling (BSEM) with informative priors and sensitivity analyses to address small-sample uncertainty. Our findings revealed a theoretically supported hierarchical structure with a strong general factor and distinct specific factors: entrepreneurial career mindset, innovativeness, motivation, and implementation, enhancing the interpretation of scores. This study guides ECCI usage by suggesting total scores for broad assessments and domain scores for diagnostic feedback. Methodologically, the findings demonstrate that combining frequentist and Bayesian approaches across samples strengthened structural validity and provided insights into evaluating imprecise responses to self-report measures and addressing sampling constraints. Overall, this work contributes a robust structural model of the ECCI and enriches the EC literature, serving as a framework for refining, testing and applying attribute-based EC measures in diverse contexts. Full article

Biological systems regulate motion and suppress unwanted vibrations through learning, adaptation, and predictive control under uncertainty. Inspired by these principles, Bayesian system identification has emerged as a powerful framework for modeling and estimation, particularly in the presence of uncertainty in structural systems. Flexible structures in aerospace and robotics require advanced control to mitigate vibrations under model uncertainty. This paper proposes a data-driven strategy leveraging a Gaussian Process (GP) integrated within a Nonlinear Model Predictive Control (NMPC) framework. The core innovation lies in using a Gaussian Process Nonlinear AutoRegressive model with eXogenous input (GP-NARX) as a probabilistic predictor to capture structural dynamics while quantifying uncertainty. The operational mechanism involves a tight coupling where the GP provides multi-step-ahead forecasts that the NMPC optimizer uses to minimize a cost function subject to constraints. Validated through simulations on Duffing oscillators, linear oscillators, and cantilever beams, the GP-NMPC achieved an 88.2% reduction in displacement amplitude compared to uncontrolled systems. Quantitative analysis shows high predictive accuracy, with a Root Mean Square Error (RMSE) of 0.0031 and a Standardized Mean-Squared Error (SMSE) below 0.05. Furthermore, Mean Standardized Log Loss (MSLL) evaluations confirm the reliability of the predictive uncertainty within the control loop. These results demonstrate strong performance in both regulation and tracking tasks, justifying this Bayesian-predictive coupling as a powerful approach for high-performance structural vibration control and a potential foundation for bio-inspired mechanical design. Full article

Residential air conditioning systems are a major contributor to household electricity consumption in tropical regions, where environmental factors such as climate variability and particulate pollution (PM10) can further increase cooling demand and accelerate equipment degradation. This study proposes an Enhanced Deep Q-Network (Enhanced DQN) for energy-efficient and maintenance-aware control of residential split-type air conditioners under dynamic environmental conditions. The proposed method integrates several stability-oriented reinforcement learning mechanisms, including Double Q-learning, a dueling architecture, prioritized experience replay, multi-step returns, Bayesian-style regularization via Monte Carlo dropout, and entropy-aware exploration. The framework is evaluated through a two-stage process consisting of a diagnostic benchmark on LunarLander-v3 to assess learning stability, followed by a realistic 365-day simulation driven by Thai weather and PM10 data. Compared with a fixed 25 °C baseline, the proposed controller reduced annual electricity consumption from 5116.22 kWh to as low as 4440.03 kWh, corresponding to a saving of 13.22%. The learned policy also exhibited environmentally adaptive behavior under high PM10 conditions, indicating maintenance-aware characteristics. These findings demonstrate that reinforcement learning can provide robust, adaptive, and sustainable control strategies for residential cooling systems in tropical environments. Full article

To advance the Sustainable Development Goals, it is essential to correct imbalances in how the benefits of urban trees are distributed across different demographic and socioeconomic groups. Environmental justice studies have frequently overlooked assumptions regarding the data-generating process and have not considered spatial confounding. This oversight potentially misestimates patterns of inequity. This study evaluates the sensitivity of inequity to model assumptions using urban tree inventories from Málaga and Sevilla and Bayesian hierarchical models. City-level differences dominated the inequity patterns, and model specification influenced the magnitude, precision, and credibility of estimated effects, though directionality remained consistent. Patterns were highly consistent across the four ecosystem services, indicating that model assumptions affected all services equivalently. Málaga and Seville exhibited divergent inequity patterns, indicating that local urban context mediates these relationships. In Seville, inequity patterns were inconsistent with the luxury hypothesis and occurred primarily across age-based demographic strata, whereas in Málaga they manifested predominantly along ethnicity, with weaker evidence of income inequities. We advocate for explicitly modeling spatial data-generating processes and comparing conventional versus confounding-mitigated approaches. This city-specific rigor is essential for urban planners to prevent resource misallocation, ensuring that tree-planting strategies address genuine inequities rather than methodological biases. Full article

Background: Human lead exposure is a multi-pathway phenomenon that integrates internal biological burden with persistent residential environmental reservoirs. Although individual lead metrics have been linked to cardiometabolic dysfunction, current research often fails to capture the ‘exposome’ reality of joint, nonlinear, and interaction-dependent effects on metabolic outcomes like BMI. Objectives: To evaluate associations between biological (blood and urinary) and residential dust (window and floor) lead measures and BMI, and to characterize nonlinear and interaction-dependent mixture effects using Bayesian Kernel Machine Regression (BKMR). Methods: We analyzed data from NHANES 2001–2002, a nationally representative survey of the U.S. noninstitutionalized civilian population. Window and floor dust lead (µg/ft²) were obtained from the NHANES household dust component, and blood lead (µg/dL) and urinary lead (µg/L) were measured using standardized NHANES laboratory protocols. BMI was calculated from measured height and weight. Missing data were addressed using multivariate imputation by chained equations. Descriptive statistics and multivariable linear regression were used to estimate adjusted associations between individual lead metrics and BMI, controlling for age, gender, income, race/ethnicity, and education. BKMR was then applied to evaluate joint mixture effects, estimate univariate and bivariate exposure–response functions, and quantify relative exposure importance using posterior inclusion probabilities (PIPs). Results: In covariate-adjusted linear regression, blood lead (β = −0.485; 95% CI: −0.566, −0.405; p < 0.001) and window dust lead (β = −0.00047; 95% CI: −0.00067, −0.00026; p < 0.001) were inversely associated with BMI, whereas floor dust lead was positively associated (β = 0.258; 95% CI: 0.209, 0.306; p < 0.001). Urinary lead was inversely but not significantly associated with BMI (β = −0.111; 95% CI: −0.235, 0.013; p = 0.079). In BKMR, blood lead was the dominant contributor, with a posterior inclusion probability (PIP; proportion of iterations in which an exposure is selected) of 1.00. Window dust lead showed modest inclusion (PIP = 0.26), whereas urinary and floor dust lead were not selected (PIP = 0.00). Exposure–response functions indicated modest nonlinearity for blood lead and greater divergence for the blood lead–window dust lead pairing at higher exposure levels. The overall mixture effect declined across increasing joint exposure quantiles, crossing the null near the median and becoming increasingly negative at higher mixture levels. Conclusions: In our study, lead metrics showed heterogeneous associations with BMI, and BKMR indicated that internal lead burden (blood lead) primarily drove mixture-related BMI patterns, with evidence that window dust lead may modify mixture effects at higher co-exposure levels. These findings support evaluating multiple lead exposure pathways jointly and using flexible mixture models to capture nonlinear and interaction-dependent relationships with BMI. Full article

Accurate short-term lightning forecasting requires reliable representations of both lightning occurrence and intensity, as well as the underlying convective processes. While ensemble prediction systems (EPSs) provide valuable probabilistic information, their ability to resolve mesoscale and convective-scale variability remains limited. In this study, we assess the added value of mesoscale information for probabilistic lightning forecasting over eastern China. A mesoscale ensemble is constructed from deterministic forecasts of the China Meteorological Administration (CMA) Mesoscale Model (MESO) using spatiotemporal neighborhood and time-lagged techniques and is combined with predictors from the CMA Regional Ensemble Prediction System (REPS). Lightning occurrence and counts are modeled within a Bayesian additive model for location, scale, and shape (BAMLSS) framework, using a hurdle-based count regression to account for excess zeros and overdispersion. Influential nonlinear predictors are selected via stability selection combined with gradient boosting. Forecast performance with and without MESO-derived predictors is systematically evaluated. The results indicate that incorporating mesoscale information generally improves forecast skill for both lightning occurrence and intensity across multiple verification metrics. These improvements are associated with MESO-derived predictors related to convective available potential energy and convective precipitation, suggesting the importance of mesoscale processes for probabilistic lightning forecasting. Full article

Accurate structural analysis is essential for the design and optimization of aircraft wings; however, repeated high-fidelity finite element analysis (FEA) becomes computationally expensive when embedded in iterative design loops. This study presents a machine learning-based surrogate modeling framework for the efficient analysis and optimization of metallic commercial wing structures. A detailed Airbus A320-like wing model was developed and analyzed in ANSYS 2023 R1 under modal, static, and eigenvalue buckling conditions. The general dimensions of the Airbus A320 wing were used only as a reference; the resulting model is a conceptual benchmark rather than a one-to-one geometric replica or a validated digital twin of a specific aircraft wing. Using Latin Hypercube Sampling, 340 high-fidelity samples were generated, with 300 samples used for training and validation and 40 retained as an independent holdout set. The proposed Pyramidal Deep Regression Network (PDRN), a deep learning-based surrogate model whose architecture is automatically tuned using Bayesian Optimization, was benchmarked against Artificial Neural Networks (ANNs), Ensemble Learning, Support Vector Regression (SVR), and Gaussian Process Regression (GPR). On the unseen test set, the PDRN achieved the best overall predictive performance, with RMS errors of 0.8% for mass, 3.1% for the first natural frequency, 11.5% for load factor, and 11.4% for safety factor. To evaluate its practical utility, the trained PDRN was embedded into a PSO-based optimization framework for mass minimization under minimum safety factor, load factor, and first-frequency constraints. The surrogate-guided optimum was verified in ANSYS and remained feasible, yielding a mass of 10,485 kg, a first natural frequency of 1.4142 Hz, a load factor of 1.307, and a safety factor of 1.158. Compared with direct ANSYS in-the-loop optimization, the proposed workflow reached a comparable feasible design with substantially fewer high-fidelity evaluations. These results demonstrate that the PDRN provides an accurate and computationally efficient surrogate for rapid wing analysis and constraint-driven structural optimization. Full article