Search Results (6,062)

Detecting small moving objects under challenging lighting conditions, such as overexposure and underexposure, remains a critical challenge in computer vision applications including surveillance, autonomous driving, and anti-UAV systems. Traditional RGB-based detectors often suffer from degraded object visibility and highly dynamic illumination, leading to suboptimal performance. To address these limitations, we propose a novel RGB-Event fusion framework that leverages the complementary strengths of RGB and event modalities for enhanced small object detection. Specifically, we introduce a Temporal Multi-Scale Attention Fusion (TMAF) module to encode motion cues from event streams at multiple temporal scales, thereby enhancing the saliency of small object features. Furthermore, we design a Sparse Noisy Gated Attention Fusion (SNGAF) module, inspired by the mixture-of-experts paradigm, which employs a sparse gating mechanism to adaptively combine multiple fusion experts based on input characteristics, enabling flexible and robust RGB-Event feature integration. Additionally, we present RGBE-UAV, which is a new RGB-Event dataset tailored for small moving object detection under diverse exposure conditions. Extensive experiments on our RGBE-UAV and public DSEC-MOD datasets demonstrate that our method outperforms existing state-of-the-art RGB-Event fusion approaches, validating its effectiveness and generalization under complex lighting conditions. Full article

Background: Differentiating pediatric pneumonia from acute bronchitis remains a persistent clinical challenge due to overlapping symptoms, often leading to diagnostic uncertainty and inappropriate antibiotic use. Methods: This study introduces DAPLEX, a structured ensemble learning framework designed to enhance diagnostic accuracy and reliability. A retrospective cohort of 868 pediatric patients was analyzed. DAPLEX was developed in three phases: (i) deployment of diverse base learners from multiple learning paradigms; (ii) multi-criteria evaluation and pruning based on generalization stability to retain a subset of well-generalized and stable learners; and (iii) complementarity-driven knowledge fusion. In the final phase, out-of-fold predicted probabilities from the retained base learners were combined with a consensus-based feature importance profile to construct a hybrid meta-input for a Multilayer Perceptron (MLP) meta-learner. Results: DAPLEX achieved a balanced accuracy of 95.3%, an F1-score of ~0.96, and a ROC-AUC of ~0.99 on an independent holdout test. Compared to the range of performance from the weakest to the strongest base learner, DAPLEX improved balanced accuracy by 3.5–5.2%, enhanced the F1-score by 4.4–5.6%, and increased sensitivity by a substantial 8.2–13.6%. Crucially, DAPLEX’s performance remained robust and consistent across all evaluated demographic subgroups, confirming its fairness and potential for broad clinical. Conclusions: The DAPLEX framework offers a robust and transparent pipeline for diagnostic decision support. By systematically integrating diverse predictive models and synthesizing both outcome predictions and key feature insights, DAPLEX substantially reduces diagnostic uncertainty in differentiating pediatric pneumonia and acute bronchitis and demonstrates strong potential for clinical application. Full article

Although epsilon-near-zero (ENZ) media have emerged as a promising platform for power dividers, the majority of existing designs are confined to fixed power splitting. In this work, two dynamically tunable power dividers using waveguide ENZ media are proposed by precisely modulating the internal magnetic field and the widths of the output waveguides. The first approach features a mechanically reconfigurable ring-shaped ENZ waveguide. By continuously re-distributing the magnetic field within the ENZ tunneling channels utilizing rotatable copper plates, arbitrary power division among multiple output ports is constructed. The second design integrates a rectangular-loop ENZ cavity into a substrate-integrated waveguide, with four positive–intrinsic–negative diodes embedded to dynamically activate specific output ports. This configuration steers electromagnetic energy toward output ports with varying cross-sectional areas, enabling on-demand control over both the power division and the number of output ports. Both analytical and full-wave simulation results confirm dynamic power division, with transmission efficiencies exceeding 93%. Despite differences in structure and actuation mechanisms, both designs exhibit flexible field control, high reconfigurability, and excellent transmission performance, highlighting their potential in advanced applications such as real-time wireless communications, multi-input–multi-output systems, and reconfigurable antennas. Full article

To foster the performance of wireless communication while saving energy, the integration of Intelligent Reflecting Surfaces (IRS) into autonomous vehicle (AV) communication networks is considered a powerful technique. This paper proposes a novel IRS-assisted vehicular communication model that combines Lagrange optimization and Gradient-Based Phase Optimization to determine the optimal transmission power, optimal interference transmission power, and IRS phase shifts. Additionally, the proposed model help increase the Signal-to-Interference-plus-Noise Ratio (SINR) by utilizing IRS, which leads to maximizes energy efficiency and the achievable data rate under a variety of environmental conditions, while guaranteeing that resource limits are satisfied. In order to represent dense vehicular environments, practical constraints for the system model, such as IRS reflection efficiency and interference, have been incorporated from multiple sources, namely, Device-to-Device (D2D), Vehicle-to-Vehicle (V2V), Vehicle-to-Base Station (V2B), and Cellular User Equipment (CUE). A Lagrangian optimization approach has been implemented to determine the required transmission interference power and the best IRS phase designs in order to enhance the system performance. Consequently, a one-dimensional convolutional neural network has been implemented for the optimized data provided by this framework as training input. This deep learning algorithm learns to predict the required optimal IRS settings quickly, allowing for real-time adaptation in dynamic wireless environments. The obtained results from the simulation show that the combined optimization and prediction strategy considerably enhances the system reliability and energy efficiency over baseline techniques. This study lays a solid foundation for implementing IRS-assisted AV networks in real-world settings, hence facilitating the development of next-generation vehicular communication systems that are both performance-driven and energy-efficient. Full article

Given the relevance and wide use of the Generative Adversarial (GA) methodology, this paper focuses on finite samples to better understand its benefits and pitfalls. We focus on its finite-sample properties from both statistical and numerical perspectives. We set up a simple and ideal “controlled experiment” where the input data are an i.i.d. Gaussian series where the mean is to be learned, and the discriminant and generator are in the same distributional family, not a neural network (NN), as in the popular GAN. We show that, even with the ideal discriminant, the classical GA methodology delivers a biased estimator while producing multiple local optima, confusing numerical methods. The situation worsens when the discriminator is in the correct parametric family but is not the oracle, leading to the absence of a saddle point. To improve the quality of the estimators within the GA method, we propose an alternative loss function, the alternative GA method, that leads to a unique saddle point with better statistical properties. Our findings are intended to start a conversation on the potential pitfalls of GA and GAN methods. In this spirit, the ideas presented here should be explored in other distributional cases and will be extended to the actual use of an NN for discriminators and generators. Full article

The Honeycomb Distribution Network is a new distribution network architecture that utilizes the Smart Power-Exchange Station (SPES) to enable power interconnection and mutual assistance among multiple microgrids/distribution units, thereby supporting high-proportion integration of distributed renewable energy and promoting a sustainable energy transition. To promote the continuous and reliable operation of the Honeycomb Distribution Network, this paper proposes a Hierarchical Switching Control Strategy to address the issues of DC bus voltage (U_dc) fluctuation in the SPES of the Honeycomb Distribution Network, as well as the state of charge (SOC) and charging/discharging power limitation of the energy storage module (ESM). The strategy consists of the system decision-making layer and the converter control layer. The system decision-making layer selects the main converter through the importance degree of each distribution unit and determines the control strategy of each converter through the operation state of the ESM’s SOC. The converter control layer restricts the ESM’s input/output active power—this ensures the ESM’s SOC and input/output active power stay within the power boundary. Additionally, it combines the Flexible Virtual Inertia Adaptive (FVIA) control method to suppress U_dc fluctuations and improve the response speed of the ESM converter’s input/output active power. A simulation model built in MATLAB/Simulink is used to verify the proposed control strategy, and the results demonstrate that the strategy can not only effectively reduce U_dc deviation and make the ESM’s input/output power reach the stable value faster, but also effectively avoid the ESM entering the unstable operation area. Full article

Despite the close linkage between carbon markets and fossil fuel markets, minimal research has investigated their co-movement dynamics during times of heightened geopolitical instability and public health crises, including the COVID-19 pandemic, Israeli–Palestinian conflict, and the Russia–Ukraine war. Most studies use conditional mean regression models for testing linear Granger causality, which falls short in assessing time-varying causal relationships. This paper employs a time-varying Granger causality framework to examine the dynamic linkages between fossil fuel markets and carbon markets across multiple time horizons. This methodology enables the evaluation of causal relationships that evolve over time, providing deeper insights into how the carbon market interacts with traditional fossil fuel markets. The study examines causal linkages among carbon, coal, and oil prices from 2 January 2018 to 11 July 2025, using data from Wind Database. The findings reveal a short-lived yet highly significant bidirectional causality between the carbon and fossil fuel markets during the COVID-19 period, whereas a sustained and highly significant bidirectional causal relationship emerges after the onset of the Russia–Ukraine war. During the outbreak of the Israeli–Palestinian conflict, this linkage continued without major disruptions or directional shifts. Furthermore, the recursive evolution approach, based on variable sub-window sizes, detects additional evidence of significant bidirectional causal relationships among carbon, coal, and oil prices. These discoveries can serve as valuable inputs for investors and policymakers, enabling them to make informed decisions that protect their interests and ensure market stability. Additionally, coal prices showed greater persistence than oil prices in these bidirectional causal links. Full article

This study illustrates the complex interaction between environmental parameters and carbonate distribution in marine sediments along the Tarfaya–Boujdour coastline (26–28° N) of Northwest Africa. Analysis of 21 surface sediment samples and their associated bottom water properties (salinity, temperature, dissolved oxygen, nutrients) reveals CaCO₃ content ranging from 16.8 wt.% to 60.5 wt.%, with concentrations above 45 wt.% occurring in multiple stations, especially in nearshore deposits. Mineralogy indicates a general decrease in quartz, with an arithmetic mean and standard deviation of 52.5 wt.% ± 19.8 towards the open sea, and an increase in carbonate minerals (calcite ≤ 24%, aragonite ≤ 10%) with depth. Sediments are predominantly composed of fine sand (78–99%), poorly classified, with gravel content reaching 6.7% in energetic coastal stations. An inverse relationship between organic carbon (0.63–3.23 wt.%) and carbonates is observed in upwelling zones, correlated with nitrate concentrations exceeding 19 μmol/L. Hydrological gradients show temperatures from 12.41 °C (offshore) to 21.62 °C (inshore), salinity from 35.64 to 36.81 psu and dissolved oxygen from 2.06 to 4.21 mL/L. The weak correlation between carbonates and depth (r = 0.10) reflects the balance between three processes: biogenic production stimulated by upwelling, dilution by Saharan terrigenous inputs, and hydrodynamic sorting redistributing bioclasts. These results underline the need for models integrating hydrology, mineralogy and hydrodynamics to predict carbonate dynamics in desert margins under upwelling. Full article

The rapid development of wireless communication technology is leading to increasingly scarce spectrum resources, making efficient utilization a critical challenge. This paper proposes a Convolutional Neural Network–Long Short-Term Memory-Integrated Gradient-Weighted Class Activation Mapping (GC-CNN-LSTM) model, aimed at enhancing the accuracy of long-term spectrum prediction across multiple frequency bands and improving model interpretability. First, we achieve multi-frequency long-term spectrum prediction using a CNN-LSTM and compare its performance against models including LSTM, GRU, CNN, Transformer, and CNN-LSTM-Attention. Next, we use an improved Grad-CAM method to explain the model and obtain global heatmaps in the time–frequency domain. Finally, based on these interpretable results, we optimize the input data by selecting high-importance frequency points and removing low-importance time segments, thereby enhancing prediction accuracy. The simulation results show that the Grad-CAM-based approach achieves good interpretability, reducing RMSE and MAPE by 6.22% and 4.25%, respectively, compared to CNN-LSTM, while a similar optimization using SHapley Additive exPlanations (SHAP) achieves reductions of 0.86% and 3.55%. Full article

Designing efficient nanoparticle-enhanced CO₂ capture systems is challenging due to the diversity of nanoparticles, solvent formulations, reactor configurations, and operating conditions. This study presents the first ANN-based meta-analysis framework developed to predict CO₂ absorption enhancement across multiple reactor systems, including batch reactors, packed columns, and membrane contactors. A curated dataset of 312 experimental data points was compiled from literature, and an artificial neural network (ANN) model was trained using six input variables: nanoparticle type, concentration, system configuration, base fluid, pressure, and temperature. The proposed model achieved high predictive accuracy (R² > 0.92; RMSE: 4.2%; MAE: 3.1%) and successfully captured complex nonlinear interactions. Feature importance analysis revealed nanoparticle concentration (28.3%) and system configuration (22.1%) as the most influential factors, with functionalized nanoparticles such as Fe₃O₄@SiO₂-NH₂ showing superior performance. The model further predicted up to 130% enhancement for ZnO in optimized membrane contactors. This AI-driven tool provides quantitative insights and a scalable decision-support framework for designing advanced nanoparticle–solvent systems, reducing experimental workload, and accelerating the development of sustainable CO₂ capture technologies. Full article

Accurate modeling of car-following behavior is essential for understanding traffic dynamics and enabling predictive control in intelligent transportation systems. This study presents a novel data-driven framework that combines information-theoretic input selection via conditional transfer entropy (CTE) with dynamic mode decomposition with control (DMDc) for identifying and forecasting car-following dynamics. In the first step, CTE is employed to identify the specific vehicles that exert directional influence on a given subject vehicle, thereby systematically determining the relevant control inputs for modeling its behavior. In the second step, DMDc is applied to estimate and predict the dynamics by reconstructing the closed-form expression of the dynamical system governing the subject vehicle’s motion. Unlike conventional machine learning models that typically seek a single generalized representation across all drivers, our framework develops individualized models that explicitly preserve driver heterogeneity. Using both synthetic data from multiple traffic models and real-world naturalistic driving datasets, we demonstrate that DMDc accurately captures nonlinear vehicle interactions and achieves high-fidelity short-term predictions. Analysis of the estimated system matrices reveals that DMDc naturally approximates kinematic relationships, further reinforcing its interpretability. Importantly, this is the first study to apply DMDc to model and predict car-following behavior using real-world driving data. The proposed framework offers a computationally efficient and interpretable tool for traffic behavior analysis, with potential applications in adaptive traffic control, autonomous vehicle planning, and human-driver modeling. Full article

Due to the local similarities between different facial expressions and the subjective influences of annotators, large-scale facial expression datasets contain significant label noise. Recognition-based noisy labels are a key challenge in the field of deep facial expression recognition (FER). Based on this, this paper proposes a simple and effective attention-based fusion consistency network (AFCN), which suppresses the impact of uncertainty and prevents deep networks from overemphasising local features. Specifically, the AFCN comprises four modules: a sample certainty analysis module, a label correction module, an attention fusion module, and a fusion consistency learning module. Among these, the sample certainty analysis module is designed to calculate the certainty of each input facial expression image; the label correction module re-labels samples with low certainty based on the model’s prediction results; the attention fusion module identifies all possible key regions of facial expressions and fuses them; the fusion consistency learning module constrains the model to maintain consistency between the regions of interest for the actual labels of facial expressions and the fusion of all possible key regions of facial expressions. This guides the model to perceive and learn global facial expression features and prevents it from incorrectly classifying expressions based solely on local features associated with noisy labels. Experiments are conducted on multiple noisy datasets to validate the effectiveness of the proposed method. The experimental results illustrate that the proposed method outperforms current state-of-the-art methods, achieving a 3.03% accuracy improvement on the 30% noisy RAF-DB dataset in particular. Full article

This paper explores the feasibility of using a wireless Inductive Resonant Link (IRL) for wearable posture monitoring. The proposed system is based on magnetically coupled textile resonators and is implemented using a Single Input Multiple Output (SIMO) configuration. In particular, the setup consists of four inductively coupled resonators: one transmitting coil integrated into a textile structure and positioned on the back of the neck, and three receiving coils placed on the shoulders. The magnetic coupling between these elements varies as a function of the user’s posture, making it possible to monitor postural changes by analyzing variations in the transmission coefficients of the link. Unlike traditional sensor-based systems that require multiple components and data processing, the proposed method uses the inherent response of the inductive link to detect posture in a simple and efficient way. To validate the concept, experimental measurements of the scattering parameters were carried out using a compact and low-power vector network analyzer. The results show a consistent and measurable relationship between postural changes and variations in the transmission coefficients, demonstrating the effectiveness of the proposed system in distinguishing between different postures. The findings suggest that inductive resonant wireless links, especially when implemented with textile components, represent a promising alternative to traditional wearable sensor technologies for posture tracking. The approach offers significant advantages in terms of wearability, power consumption, and simplicity, making it suitable for applications in ergonomics, rehabilitation, occupational health, and smart clothing. Full article

With the booming development of the low-altitude economy and the widespread application of Unmanned Aerial Vehicles (UAVs), integrated sensing and communication (ISAC) technology plays an increasingly pivotal role in intelligent communication networks. However, low-altitude platforms supporting ISAC, such as UAV swarms, are highly vulnerable to deception jamming in complex electromagnetic environments. Existing multistatic radar systems face challenges in processing slowly fluctuating targets (like low-altitude UAVs) and adapting to complex electromagnetic environments when fusing multiple pulse echoes. To address this issue, targeting the protection needs of low-altitude targets like UAVs, this paper leverages the characteristic of rapid amplitude fluctuation in frequency-agile radar echoes to analyze the differences between true and false targets in multistatic frequency-agile radar systems, particularly for slowly fluctuating UAV targets, demonstrating the feasibility of discrimination. Building on this, we introduce a neural network approach to deeply extract discriminative features from true and false target echoes and propose a neural network-based variable-length processing method for deception jamming discrimination in multistatic frequency-agile radar. The simulation results show that the proposed method effectively exploits deep-level echo features, significantly improving the discrimination probability between true and false targets, especially for slowly fluctuating UAV targets. Crucially, even when trained on a fixed number of pulses, the model can process input data with varying pulse counts, greatly enhancing its practical deployment capability in dynamic UAV mission scenarios. Full article

Understanding spatial variability in crop yields across fields is critical for developing precision agricultural strategies that optimize productivity while reducing negative environmental impacts. This variability often arises from a complex interplay of topographic features, soil characteristics, and hydrological conditions. This study investigates the influence of hydro-topographic factors on corn and soybean yield variability from 2016 to 2023 at the well-managed experimental sites in Beltsville, Maryland. A high-resolution surface digital elevation model (DEM) and subsurface DEM derived from ground-penetrating radar (GPR) were used to quantify topographic factors (elevation, slope, and aspect) and hydrological factors (surface flow accumulation, depth from the surface to the subsurface-restricting layer, and distance from each crop pixel to the nearest subsurface flow pathway). Topographic variables alone explained yield variation, with a relative root mean square error (RRMSE) of 23.7% (r² = 0.38). Adding hydrological variables reduced the error to 15.3% (r² = 0.73), and further combining with remote sensing data improved the explanatory power to an RRMSE of 10.0% (r² = 0.87). Notably, even without subsurface data, incorporating surface-derived flow accumulation reduced the RRMSE to 18.4% (r² = 0.62), which is especially important for large-scale cropland applications where subsurface data are often unavailable. Annual spatial yield variation maps were generated using hydro-topographic variables, enabling the identification of long-term persistent yield regions (LTRs), which served as stable references to reduce spatial anomalies and enhance model robustness. In addition, by combining remote sensing data with interannual meteorological variables, prediction models were evaluated with and without hydro-topographic inputs. The inclusion of hydro-topographic variables improved spatial characterization and enhanced prediction accuracy, reducing error by an average of 4.5% across multiple model combinations. These findings highlight the critical role of hydro-topography in explaining spatial yield variation for corn and soybean and support the development of precise, site-specific management strategies to enhance productivity and resource efficiency. Full article