Search Results (1,138)

Accurate modeling of ionospheric variability is critical for space weather forecasting and GNSS applications. While machine learning approaches have shown promise, progress is hindered by the absence of standardized benchmarking practices and narrow test periods. In this paper, we take the first step toward fostering rigorous and reproducible evaluation of AI models for ionospheric forecasting by introducing IonoBench—a benchmarking framework that employs a stratified data split, balancing solar intensity across subsets while preserving 16 high-impact geomagnetic storms (Dst ≤ $-100$ nT) for targeted stress testing. Using this framework, we benchmark a field-specific model (DCNN) against state-of-the-art spatiotemporal architectures (SwinLSTM and SimVPv2) using the climatological IRI 2020 model as a baseline reference. DCNN, though effective under quiet conditions, exhibits significant degradation during elevated solar and storm activity. SimVPv2 consistently provides the best performance, with superior evaluation metrics and stable error distributions. Compared to the C1PG baseline (the CODE 1-day forecast product), SimVPv2 achieves a notable RMSE reduction up to $32.1\%$ across various subsets under diverse solar conditions. The reported results highlight the value of cross-domain architectural transfer and comprehensive evaluation frameworks in ionospheric modeling. With IonoBench, we aim to provide an open-source foundation for reproducible comparisons, supporting more meticulous model evaluation and helping to bridge the gap between ionospheric research and modern spatiotemporal deep learning. Full article

In order to meet the needs for the rapid detection of crop growth and support variable management in farmland, an intelligent chlorophyll content in cotton leaves (CCC) detection system based on hyperspectral imaging (HSI) technology was designed and developed. The system includes a near-infrared (NIR) hyperspectral image acquisition module, a spectral extraction module, a main control processor module, a model acceleration module, a display module, and a power module, which are used to achieve rapid and non-destructive detection of chlorophyll content. Firstly, spectral images of cotton canopy leaves during the seedling, budding, and flowering-boll stages were collected, and the dataset was optimized using the first-order differential algorithm (1D) and Savitzky–Golay five-term quadratic smoothing (SG) algorithm. The results showed that SG had better processing performance. Secondly, the sparrow search algorithm optimized backpropagation neural network (SSA-BPNN) and one-dimensional convolutional neural network (1DCNN) algorithms were selected to establish a chlorophyll content detection model. The results showed that the determination coefficients $R_p^2$ of the chlorophyll SG-1DCNN detection model during the seedling, budding, and flowering-boll stages were 0.92, 0.97, and 0.95, respectively, and the model performance was superior to SG-SSA-BPNN. Therefore, the SG-1DCNN model was embedded into the detection system. Finally, a CCC intelligent detection system was developed using Python 3.12.3, MATLAB 2020b, and ENVI, and the system was subjected to application testing. The results showed that the average detection accuracy of the CCC intelligent detection system in the three stages was 98.522%, 99.132%, and 97.449%, respectively. Meanwhile, the average detection time for the samples is only 20.12 s. The research results can effectively solve the problem of detecting the nutritional status of cotton in the field environment, meet the real-time detection needs of the field environment, and provide solutions and technical support for the intelligent perception of crop production. Full article

Traditional bathymetry measures require a large number of human hours, and many bathymetry records are obsolete or missing. Automated measures of bathymetry would reduce costs and increase accessibility for research and applications. In this paper, we optimized a recent machine learning model, named CatBoostOpt, to estimate bathymetry based on high-resolution WorldView-2 (WV-2) multi-spectral optical satellite images. CatBoostOpt was demonstrated across the Florida Big Bend coastline, where the model learned correlations between in situ sound Navigation and Ranging (Sonar) bathymetry measurements and the corresponding multi-spectral reflectance values in WV-2 images to map bathymetry. We evaluated three different feature transformations as inputs for bathymetry estimation, including raw reflectance, log-linear, and log-ratio transforms of the raw reflectance value in WV-2 images. In addition, we investigated the contribution of each spectral band and found that utilizing all eight spectral bands in WV-2 images offers the best solution for handling complex water quality conditions. We compared CatBoostOpt with linear regression (LR), support vector machine (SVM), random forest (RF), AdaBoost, gradient boosting, and deep convolutional neural network (DCNN). CatBoostOpt with log-ratio transformed reflectance achieved the best performance with an average root mean square error (RMSE) of 0.34 and coefficient of determination (${\mathrm{R}}^2$) of 0.87. Full article

Dark web traffic classification is an important research direction in cybersecurity; however, traditional classification methods have many limitations. Although deep learning architectures like CNN and LSTM, as well as multi-structural fusion frameworks, have demonstrated partial success, they remain constrained by shallow feature representation, localized decision boundaries, and poor generalization capacity. To improve the prediction accuracy and classification precision of dark web traffic, we propose a novel dark web traffic classification model integrating multi-channel image deep learning and a three-dimensional convolutional neural network (3D-CNN). The proposed framework leverages spatial–temporal feature fusion to enhance discriminative capability, while the 3D-CNN structure effectively captures complex traffic patterns across multiple dimensions. The experimental results show that compared to common 2D-CNN and 1D-CNN classification models, the dark web traffic classification method based on multi-channel image visual features and 3D-CNN can improve classification by 5.1% and 3.3% while maintaining a smaller total number of parameters and feature recognition parameters, effectively reducing the computational complexity of the model. In comparative experiments, 3D-CNN validates the model’s superiority in accuracy and computational efficiency compared to state-of-the-art methods, offering a promising solution for dark web traffic monitoring and security applications. Full article

The accurate prediction of soil organic matter (SOM) content is essential for promoting sustainable soil management and addressing global climate change. Due to multiple factors such as topography and climate, especially in mountainous areas, SOM spatial prediction faces significant challenges. The main novelty of this study lies in proposing a geographic positional encoding mechanism that embeds geographic location information into the feature representation of a Transformer model. The encoder structure is further modified to enhance spatial awareness, resulting in the development of the Geo-Positional Transformer (GPTransformer). Furthermore, this model is integrated with a 1D-CNN to form a dual-branch neural network called the Geo-Positional Transformer-CNN (GPTransCNN). This study collected 1490 topsoil samples (0–20 cm) from cultivated land in Longshan County to develop a predictive model for mapping the spatial distribution of SOM across the entire cultivated area. Different models were comprehensively evaluated through ten-fold cross-validation, ablation experiments, and uncertainty analysis. The results show that GPTransCNN has the best performance, with an R² improvement of approximately 43% over the Transformer, 19% over the GPTransformer, and 15% over the 1D-CNN. This study demonstrates that by incorporating geographic positional information, GPTransCNN effectively combines the global modeling capabilities of the GPTransformer with the local feature extraction strengths of the 1D-CNN, which can improve the accuracy of SOM mapping in mountainous areas. This approach provides data support for sustainable soil management and decision-making in response to global climate change. Full article

Monitoring Peripheral Oxygen Saturation (SpO₂) is an important vital sign both in Intensive Care Units (ICUs), during surgery and convalescence, and as part of remote medical consultations after of the COVID-19 pandemic. This has made the development of new SpO₂-measurement tools an area of active research and opportunity. In this paper, we present a new Deep Learning (DL) combined strategy to estimate SpO₂ without contact, using pre-magnified facial videos to reveal subtle color changes related to blood flow and with no calibration per subject required. We applied the Eulerian Video Magnification technique using the Hermite Transform (EVM-HT) as a feature detector to feed a Three-Dimensional Convolutional Neural Network (3D-CNN). Additionally, parameters and hyperparameter Bayesian optimization and an ensemble technique over the dataset magnified were applied. We tested the method on 18 healthy subjects, where facial videos of the subjects, including the automatic detection of the reference from a contact pulse oximeter device, were acquired. As performance metrics for the SpO₂-estimation proposal, we calculated the Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and other parameters from the Bland–Altman (BA) analysis with respect to the reference. Therefore, a significant improvement was observed by adding the ensemble technique with respect to the only optimization, obtaining 14.32% in RMSE (reduction from 0.6204 to 0.5315) and 13.23% in MAE (reduction from 0.4323 to 0.3751). On the other hand, regarding Bland–Altman analysis, the upper and lower limits of agreement for the Mean of Differences (MOD) between the estimation and the ground truth were 1.04 and −1.05, with an MOD (bias) of −0.00175; therefore, MOD $\pm 1.96\sigma$ = −0.00175 ± 1.04. Thus, by leveraging Bayesian optimization for hyperparameter tuning and integrating a Bagging Ensemble, we achieved a significant reduction in the training error (bias), achieving a better generalization over the test set, and reducing the variance in comparison with the baseline model for SpO₂ estimation. Full article

Soils serve as critical carbon reservoirs, playing an essential role in climate change mitigation and agricultural sustainability. Accurate soil property determination relies on soil spectral reflectance data from Earth observation (EO), but current vegetation models often oversimplify soil conditions. This study introduces a novel approach that combines radiative transfer models (RTMs) with open-access soil spectral libraries to address this challenge. Focusing on conditions of low soil moisture content (SMC), photosynthetic vegetation (PV), and non-photosynthetic vegetation (NPV), the coupled Marmit–Leaf–Canopy (MLC) model is used to simulate early crop growth stages. The MLC model, which integrates MARMIT and PRO4SAIL2, enables the generation of mixed soil–vegetation scenarios. A simulated EO disturbed soil spectral library (DSSL) was created, significantly expanding the EU LUCAS cropland soil spectral library. A 1D convolutional neural network (1D-CNN) was trained on this database to predict Soil Organic Carbon (SOC) content. The results demonstrated relatively high SOC prediction accuracy compared to previous approaches that rely only on RTMs and/or machine learning approaches. Incorporating soil moisture content significantly improved performance over bare soil alone, yielding an R² of 0.86 and RMSE of 4.05 g/kg, compared to R² = 0.71 and RMSE = 6.01 g/kg for bare soil. Adding PV slightly reduced accuracy (R² = 0.71, RMSE = 6.31 g/kg), while the inclusion of NPV alongside moisture led to modest improvement (R² = 0.74, RMSE = 5.84 g/kg). The most comprehensive model, incorporating bare soil, SMC, PV, and NPV, achieved a balanced performance (R² = 0.76, RMSE = 5.49 g/kg), highlighting the importance of accounting for all surface components in SOC estimation. While further validation with additional scenarios and SOC prediction methods is needed, these findings demonstrate, for the first time, using radiative-transfer simulations of mixed vegetation-soil-water environments, that an EO-DSSL approach enhances machine learning-based SOC modeling from EO data, improving SOC mapping accuracy. This innovative framework could significantly improve global-scale SOC predictions, supporting the design of next-generation EO products for more accurate carbon monitoring. Full article

This study introduces a deep learning (DL)-based optimal condition monitoring and control system tailored to indoor smart greenhouses, with a novel focus on maintaining symmetry—defined as a dynamic equilibrium among temperature, humidity, and CO₂ levels—critical in plant growth. A hydroponic greenhouse prototype was developed to capture real-time climate data at high temporal resolution. A custom 1D convolutional neural network (1D-CNN) optimized via a genetic algorithm (GA) was employed to predict environmental fluctuations, achieving R² scores up to 0.99 and a standard error of prediction (SEP) as low as 0.35%. The system then actuated climate control mechanisms to restore and maintain symmetry. Experimental validation revealed that plants grown under the symmetry-aware control system exhibited significantly improved growth metrics. The results underscore the potential of integrating symmetry-aware DL strategies into precision agriculture in achieving sustainable and resilient plant production systems. Full article

Traditionally, Helicobacter pylori (HP) gastritis has been diagnosed by pathologists through the examination of gastric biopsies using optical microscopy with standard hematoxylin and eosin (H&E) staining. However, with the adoption of digital pathology, the identification of HP faces certain limitations, particularly due to insufficient resolution in some scanned images. Moreover, interobserver variability has been well documented in the traditional diagnostic approach, which may further complicate consistent interpretation. In this context, deep convolutional neural network (DCNN) models are showing promising results in the automated detection of this infection in whole-slide images (WSIs). The aim of the present article is to detect the presence of HP infection from our own institutional dataset of histopathological gastric biopsy samples using different pretrained and recognized DCNN and AutoML approaches. The dataset comprises 100 H&E-stained WSIs of gastric biopsies. HP infection was confirmed previously using immunohistochemical confirmation. A total of 45,795 patches were selected for model development. InceptionV3, Resnet50, and VGG16 achieved AUC (area under the curve) values of 1. However, InceptionV3 showed superior metrics such as accuracy (97%), recall (100%), F1 score (97%), and MCC (93%). BoostedNet and AutoKeras achieved accuracy, precision, recall, specificity, and F1 scores less than 85%. The InceptionV3 model was used for external validation, and the predictions across all patches yielded a global accuracy of 78%. In conclusion, DCNN models showed stronger potential for diagnosing HP in gastric biopsies compared with the auto ML approach. However, due to variability across pathology applications, no single model is universally optimal. A problem-specific approach is essential. With growing WSI adoption, DL can improve diagnostic accuracy, reduce variability, and streamline pathology workflows using automation. Full article

The application of deep learning to seismic electric signal (SES) anomaly detection remains underexplored in geophysics. This study introduces the integration of a 1D convolutional neural network (1DCNN) with a self-attention mechanism to automate SES analysis in a station in a certain place in China. Utilizing physics-informed data augmentation, our framework adapts to real-world interference scenarios, including subway operations and tidal fluctuations. The model achieves an F1-score of 0.9797 on a 7-year dataset, demonstrating superior robustness and precision compared to traditional manual interpretation. This work establishes a practical deep learning solution for real-time geoelectric anomaly monitoring, offering a transformative tool for earthquake early warning systems. Full article

With the deepening adoption of whole-lifecycle management paradigms in aviation equipment, accurate prediction of remaining useful life has emerged as a pivotal technical enabler for ensuring flight safety and optimizing maintenance resource allocation. This study systematically addresses limitations in existing data-driven remaining useful life prediction methodologies through comprehensive optimizations spanning data preprocessing protocols and model architectural enhancements. To mitigate the local optimum entrapment inherent in deep learning hyperparameter optimization, an improved Gray Wolf Optimizer incorporating dynamic perturbation factors is proposed. This algorithm is subsequently deployed to optimize hyperparameters within the redesigned predictive architecture. Comparative analyses reveal that the proposed framework achieves superior prediction accuracy compared to mainstream optimization-driven models and state-of-the-art approaches. The results substantiate the capability of dynamic perturbation strategies to enhance both hyperparameter optimization quality and prediction stability, ultimately delivering an efficient solution for aero-engine remaining useful life estimation. The experiments on the C-MAPSS Dataset verify the effectiveness of these improvements. Full article

Non-active current in the time domain is considered for application to the diagnostics and classification of loads in power grids based on waveform-distortion characteristics, taking as a working example several recordings of the pantograph current in an AC railway system. Data are processed with a deep autoencoder for feature extraction and then clustered via k-means to allow identification of patterns in the latent space. Clustering enables the evaluation of the relationship between the physical meaning and operation of the system and the distortion phenomena emerging in the waveforms during operation. Euclidean distance (ED) is used to measure the diversity and pertinence of observations within pattern groups and to identify anomalies (abnormal distortion, transients, …). This approach allows the classification of new data by assigning data to clusters based on proximity to centroids. This unsupervised method exploiting non-active current is novel and has proven useful for providing data with labels for later supervised learning performed with the 1D-CNN, which achieved a balanced accuracy of 96.46% under normal conditions. ED and 1D-CNN methods were tested on an additional unlabeled dataset and achieved 89.56% agreement in identifying normal states. Additionally, Grad-CAM, when applied to the 1D-CNN, quantitatively identifies the waveform parts that influence the model predictions, significantly enhancing the interpretability of the classification results. This is particularly useful for obtaining a better understanding of load operation, including anomalies that affect grid stability and energy efficiency. Finally, the method has been also successfully further validated for general applicability with data from a different scenario (charging of electric vehicles). The method can be applied to load identification and classification for non-intrusive load monitoring, with the aim of implementing automatic and unsupervised assessment of load behavior, including transient detection, power-quality issues and improvement in energy efficiency. Full article

Mental health disparities among those who self-identify as gay men in Peru remain a pressing public health concern, yet predictive models for early identification remain limited. This research aims to (1) develop machine learning and deep learning models to predict mental health issues in those who self-identify as gay men, and (2) evaluate the influence of demographic, economic, health-related, behavioral and social factors using interpretability techniques to enhance understanding of the factors shaping mental health outcomes. A dataset of 2186 gay men from the First Virtual Survey for LGBTIQ+ People in Peru (2017) was analyzed, considering demographic, economic, health-related, behavioral, and social factors. Several classification models were developed and compared, including Logistic Regression, Artificial Neural Networks, Random Forest, Gradient Boosting Machines, eXtreme Gradient Boosting, and a One-dimensional Convolutional Neural Network (1D-CNN). Additionally, the Shapley values and Layer-wise Relevance Propagation (LRP) heatmaps methods were used to evaluate the influence of the studied variables on the prediction of mental health issues. The results revealed that the 1D-CNN model demonstrated the strongest performance, achieving the highest classification accuracy and discrimination capability. Explainability analyses underlined prior infectious diseases diagnosis, access to medical assistance, experiences of discrimination, age, and sexual identity expression as key predictors of mental health outcomes. These findings suggest that advanced predictive techniques can provide valuable insights for identifying at-risk individuals, informing targeted interventions, and improving access to mental health care. Future research should refine these models to enhance predictive accuracy, broaden applicability, and support the integration of artificial intelligence into public health strategies aimed at addressing the mental health needs of this population. Full article

Recent advancements in deep learning have driven the rapid proliferation of deepfake generation techniques, raising substantial concerns over digital security and trustworthiness. Most current detection methods primarily focus on spatial or frequency domain features but show limited effectiveness when dealing with compressed videos and cross-dataset scenarios. Observing that mainstream generation methods use frame-by-frame synthesis without adequate temporal consistency constraints, we introduce the Spatiotemporal Attention 3D Network (STA-3D), a novel framework that combines a lightweight spatiotemporal attention module with a 3D convolutional architecture to improve detection robustness. The proposed attention module adopts a symmetric multi-branch architecture, where each branch follows a nearly identical processing pipeline to separately model temporal-channel, temporal-spatial, and intra-spatial correlations. Our framework additionally implements Spatial Pyramid Pooling (SPP) layers along the temporal axis, enabling adaptive modeling regardless of input video length. Furthermore, we mitigate the inherent asymmetry in the quantity of authentic and forged samples by replacing standard cross entropy with focal loss for training. This integration facilitates the simultaneous exploitation of inter-frame temporal discontinuities and intra-frame spatial artifacts, achieving competitive performance across various benchmark datasets under different compression conditions: for the intra-dataset setting on FF++, it improves the average accuracy by 1.09 percentage points compared to existing SOTA, with a more significant gain of 1.63 percentage points under the most challenging C40 compression level (particularly for NeuralTextures, achieving an improvement of 4.05 percentage points); while for the intra-dataset setting, AUC is enhanced by 0.24 percentage points on the DFDC-P dataset. Full article

To address the limitations in time–frequency feature representation of shearer arm gear faults and the issues of parameter redundancy and low training efficiency in standard convolutional neural networks (CNNs), this study proposes a diagnostic method based on an improved S-transform and a Depthwise Separable Convolutional Neural Network (DSCNN). First, the improved S-transform is employed to perform time–frequency analysis on the vibration signals, converting the original one-dimensional signals into two-dimensional time–frequency images to fully preserve the fault characteristics of the gear. Then, a neural network model combining standard convolution and depthwise separable convolution is constructed for fault identification. The experimental dataset includes five gear conditions: tooth deficiency, tooth breakage, tooth wear, tooth crack, and normal. The performance of various frequency-domain and time-frequency methods—Wavelet Transform, Fourier Transform, S-transform, and Gramian Angular Field (GAF)—is compared using the same network model. Furthermore, Grad-CAM is applied to visualize the responses of key convolutional layers, highlighting the regions of interest related to gear fault features. Finally, four typical CNN architectures are analyzed and compared: Deep Convolutional Neural Network (DCNN), InceptionV3, Residual Network (ResNet), and Pyramid Convolutional Neural Network (PCNN). Experimental results demonstrate that frequency–domain representations consistently outperform raw time-domain signals in fault diagnosis tasks. Grad-CAM effectively verifies the model’s accurate focus on critical fault features. Moreover, the proposed method achieves high classification accuracy while reducing both training time and the number of model parameters. Full article