Search Results (345)

Vigna umbellate, a typical edible and medicinal crop, is rich in polyphenolic compounds with antioxidant, antibacterial, anti-inflammatory, and lipid-regulating activities. However, traditional methods for polyphenol content detection rely on chemical analysis, which is cumbersome and time-consuming, making it difficult to meet the demands of high-throughput rapid detection. Although hyperspectral imaging technology offers the potential for non-destructive and rapid detection, existing analytical methods are often limited by issues such as high spectral band redundancy, insufficient feature extraction, and inadequate model stability, which constrain prediction accuracy and practical application potential. To address this, this study proposes a multi-scale residual convolutional neural network (MS-RCNN) based on competitive adaptive reweighted sampling (CARS) for feature band selection, combined with near-infrared hyperspectral imaging technology, to construct a rapid and non-destructive prediction model for the polyphenol content of Vigna umbellata. The model employs a parallel multi-scale convolutional module to extract spectral features with different receptive fields, and incorporates residual connections and adaptive pooling mechanisms to enhance feature reuse and robustness. Experiments compared the performance of partial least squares regression (PLSR), least squares support vector machine (LS-SVM), multi-scale convolutional neural network (MS-CNN), and MS-RCNN models. The results indicate that the MS-RCNN model based on CARS screening achieved the best prediction performance, with a coefficient of determination (R²) of 0.9467, a root mean square error of prediction (RMSEP) of 0.0448, and a residual predictive deviation (RPD) of 4.33. Compared with the optimal PLSR and LSSVM models, its R² values were improved by 0.2078 and 0.1119, respectively. In summary, the MS-RCNN model proposed in this study enables rapid, non-destructive, and accurate prediction of polyphenol content in Vigna umbellata, providing an efficient technical approach for quality detection of edible and medicinal crops. Full article

Accurate tail risk forecasting in emerging markets is frequently compromised by the nonlinear dynamics and time-varying long memory of high-frequency volatility. In this study, we employ multifractal detrended fluctuation analysis (MF-DFA) to decode the complex market behavior, revealing pronounced multifractality and strong persistence that defy the static assumptions of classical linear models. The multifractal analysis is only used for research motivation and model design, not as input features for the model. To bridge the gap between fractal diagnostics and predictive modeling, we propose an attention-based dynamically reweighted SA-HAR-J-Net framework. This architecture uniquely integrates HAR-style multi-horizon inputs with a bidirectional LSTM (BiLSTM) encoder and a temporal self-attention mechanism. Crucially, the attention module functions as a dynamic reweighting system, allowing the model to adaptively emphasize historical patterns that receive higher attention weights under changing market conditions, thereby mimicking the time-varying correlations inherent in multifractal processes. Furthermore, we incorporate jump proxies and realized higher moments to enhance the capture of extreme tail dynamics. Utilizing a strict expanding-window out-of-sample protocol, the proposed method achieves significantly lower quantile loss and superior calibration relative to established econometric and machine learning benchmarks for Value-at-Risk (VaR) forecasting. This work provides a robust framework for tail risk monitoring by effectively aligning deep learning architectures with the stylized facts of multifractal markets. Full article

Unsupervised domain adaptation (UDA) has been extensively studied for bearing fault diagnosis under multiple operating conditions by mitigating distribution discrepancies across domains. However, in cross-domain imbalanced scenarios, bearing vibration signals are affected by both feature shift and class imbalance. Although a robust decision boundary learned from the source domain is critical for reliable transfer, classifier discriminability and robustness can be degraded by hard samples located near the boundary. As a result, the decision boundary may become ambiguous during adaptation, leading to degraded diagnostic performance in the target domain. To address these issues, a Maximum Margin Local Domain Adaptation (MMLDA) framework is proposed in which a multi-scale convolutional neural network is adopted as the backbone. Three core components are integrated into our framework: first, category-level reweighting to alleviate source-domain class imbalance; second, cross-domain local category alignment to reduce fine-grained feature discrepancies and feature shift; and finally, maximum-margin loss regularization to impose adaptive margin constraints on hard samples for improved decision boundary robustness. To evaluate the proposed method, cross-domain imbalanced transfer tasks under multiple operating conditions were constructed on two public bearing fault datasets, and comparative experiments were conducted. The results under different imbalance protocols demonstrate improved robustness and generalization of MMLDA. Full article

Industrial equipment data often exhibit high dimensionality and class imbalance, which make it difficult to achieve both accurate failure detection and identification of the factors contributing to failures. To address this issue, this study proposes an explainable failure classification framework, Feature-Wise Attention Graph Convolutional Network (FWA-GCN), which combines Feature-Wise Attention (FWA) with a Graph Convolutional Network (GCN) to provide both high classification performance and variable-level interpretability. In the proposed model, tabular sensor records are treated as nodes, and a similarity-based graph is constructed to capture relationships among samples. Feature-Wise Attention learns the importance of each feature and reweights node features accordingly, and the reweighted features are then used as input to the GCN to classify failure occurrences. To alleviate the class imbalance problem, a weighted loss function is applied during training by assigning a higher weight to the failure class. Experiments conducted on the Air Pressure System (APS) dataset demonstrate that the proposed FWA-GCN achieves Precision of 79.95%, Recall of 85.07%, and F1-score of 82.43%, outperforming conventional machine learning models including Random Forest, XGBoost, CatBoost, and Multi-Layer Perceptron, as well as a standard GCN model. Furthermore, an ablation study was conducted by removing the top features selected by the attention mechanism. The results show a significant decrease in recall, confirming the effectiveness of the attention-based feature importance and supporting the interpretability of the proposed framework. Full article

Accurate forecasting of mining-induced surface deformation is critical for coal-mine safety assessment and hazard mitigation. InSAR deformation time series are often short, temporally sparse, and strongly nonlinear. These characteristics can make purely data-driven predictors unreliable in small-sample settings. To address this issue, we propose a generation–prediction–interpretation framework that combines generative augmentation with physics-informed forecasting. We first develop a TCN-TimeGAN model to synthesize high-fidelity deformation sequences and expand the training set. Recurrent modules in the generator and discriminator are replaced with causal TCN residual blocks, and a temporal self-attention layer is further stacked on top of the TCN backbone to adaptively reweight informative time steps. We then construct a physics-informed Kolmogorov–Arnold Network, termed PI-KAN. Subsidence-consistency and smoothness priors are embedded in the learning objective to promote physically plausible predictions while retaining spline-based interpretability. Experiments on SBAS-InSAR deformation series from the Guqiao coal mine show that the framework achieves an RMSE of 0.825 mm and an R² of 0.968. It outperforms TGAN-KAN, CNN-BiGRU, and BiGRU under the same evaluation protocol. Visualizations of the learned spline-based edge functions further reveal stronger nonlinear responses for lagged inputs closer to the forecast horizon, providing interpretable evidence of short-term temporal sensitivity under sparse observations. Full article

Early-stage fungal contamination in maize kernels is difficult to identify visually and it can cause severe quality and safety risks during storage and transportation. Short-wave infrared (SWIR) hyperspectral imaging offers a rapid, non-destructive approach by capturing chemical information related to water, proteins, and lipids. This study investigates the early detection and classification of Gibberella zeae contamination in maize kernels using SWIR hyperspectral imaging combined with machine learning. Two maize varieties were artificially inoculated and cultured under controlled conditions, followed by hyperspectral data collection over six contamination stages. Various preprocessing techniques including standard normal variate (SNV), second derivative (SD), multiplicative scatter correction (MSC), and derivatives were evaluated to enhance data quality. Feature wavelength selection was performed using successive projections algorithm (SPA), competitive adaptive reweighted sampling (CARS), and uninformative variable elimination (UVE), significantly reducing redundancy and improving classification performance. Multiple models, including linear discriminant analysis (LDA), multilayer perceptron (MLP), support vector machine (SVM), a convolutional neural network (CNN), long short-term memory (LSTM) network, and a hybrid architecture Transformer that integrated a CNN, a LSTM network, and a Transformer (abbreviated as CLT), were constructed for both binary (healthy vs. contaminated) and multiclass classification tasks. Specifically, the multiclass task consisted of six contamination stages corresponding to contamination time from Day 0 to Day 5. The best binary classification task accuracy of 100% was achieved using SNV-preprocessed data with the MLP model. For multiclass classification task, the SD-preprocessed LDA model reached a test accuracy of 92.56%. Combined with appropriate preprocessing, feature selection and modeling, these results demonstrate that hyperspectral imaging is a powerful tool for the non-destructive, early-stage identification of fungal contamination in maize kernels, offering strong support for food safety and quality monitoring. Full article

Nitrogen is one of the essential nutrient elements that affect rice growth, yield, and quality formation. Accurate and timely estimation of rice nitrogen status is fundamental for precision fertilization in agricultural fields. Hyperspectral remote sensing technology provides a promising approach for rapid and accurate acquisition of nitrogen status of rice in the field. However, traditional single-angle hyperspectral observations are easily disturbed by factors such as canopy structure, light direction, and background reflection, limiting their inversion accuracy and stability. This study is based on multi-angle polarimetric hyperspectral data obtained from an unmanned aerial vehicle platform. It extracts features from multi-angle polarimetric spectra based on three algorithms: successive projections algorithm (SPA), competitive adaptive reweighted sampling, and relevant features. The input weight and hidden layer bias of the extreme learning machine (ELM) model were optimized by the whale optimization algorithm (WOA) and caterpillar fungus optimization algorithm (CFO), taking the sensitive band of optimal viewing angle as input. Finally, an inversion model of rice canopy nitrogen content (CNC) based on multi-angle polarization hyperspectral data was established. The results demonstrate that the inversion results of the combination of SPA-(30°) + SPA-(45°) observation angles and feature selection methods are optimal, and multi-angle fusion significantly improves the model’s ability to characterize CNC, with higher stability and accuracy than single-angle modeling. The R² of CFO-ELM on the training set and test set reach 0.8553 and 0.8274, respectively, which is significantly better than the original ELM and WOA-ELM, becoming the optimal CNC inversion model in this study. The rice CNC inversion model based on multi-angle polarimetric hyperspectral data constructed in this study provides a specific reference for the rapid detection of rice CNC. Full article

Monocular 3D reconstruction remains a persistent challenge for autonomous driving systems in Degraded Visual Environments (DVEs) with extreme glare and low illumination, such as highway tunnels, due to the lack of reliable texture cues. This paper proposes a physics-aware deep learning framework that overcomes these limitations by fusing polarization sensing with conventional intensity imaging. Unlike traditional end-to-end data-driven fusion strategies, we propose a Modality-Aligned Parameter Injectionstrategy. By remapping the weight space of the input layer, this strategy achieves a smooth transfer of the pre-trained Vision Transformer (i.e., MiDaS) to multi-modal inputs. Its core advantage lies in the seamless integration of four-channel polarization geometric information while fully preserving the pre-trained semantic representation capabilities of the backbone network, thereby avoiding the overfitting risk associated with training from scratch on small-sample data. Furthermore, we design a Reliability-Aware Gating mechanism that dynamically re-weights appearance and geometric cues based on intensity saturation and the physical validity of polarization signals as measured by the Degree of Linear Polarization (DoLP). We validate the proposed method on our self-constructed POLAR-GLV benchmark, a real-world dataset collected specifically for high dynamic range tunnel scenarios. Extensive experiments demonstrate that our method consistently outperforms intensity-only baselines, reducing geometric reconstruction error by 24.2% in high-glare tunnel exit zones and 10.0% at tunnel entrances. Crucially, compared to multi-stream fusion architectures, these performance gains come with negligible additional computational cost, making the framework highly suitable for resource-constrained onboard inference environments. Full article

This study presents a genome-resolved shotgun metagenomic analysis of artisanal raw-milk goat cheese from the Masurian region of Poland, addressing the limited understanding of strain-level diversification and functional restructuring during traditional cheese ripening. While microbial succession in cheese has been widely described, comprehensive genome-resolved analyses integrating strain-level genomic heterogeneity, pathway reweighting, and mobile genetic elements in artisanal goat cheese remain scarce. By combining taxonomic profiling with metagenome-assembled genome (MAG) reconstruction and pathway-level functional analysis, we characterised microbial succession and genome plasticity across ripening stages. Genome reconstruction yielded 37 MAGs during early ripening and 141 MAGs in mature cheese, revealing increased genome recoverability and pronounced strain-level heterogeneity within dominant taxa, including Lactiplantibacillus plantarum, Lacticaseibacillus paracasei, and Lactococcus lactis. Alpha diversity increased in mature samples, consistent with progressive community restructuring. Functional profiling demonstrated coordinated metabolic reweighting, particularly within carbohydrate metabolism, while amino acid and lipid metabolism remained proportionally stable. Genome-resolved analyses further identified tetracycline- and sulfonamide-associated resistance determinants and diverse bacteriophages targeting lactic acid bacteria, highlighting the role of mobile genetic elements in horizontal gene transfer and microevolutionary adaptation during ripening. Full article

Background/Objectives: Automatic segmentation and identification of vertebrae in spinal CT are essential for assisting diagnosis of spinal disorders and for preoperative planning. The task is challenging due to the high structural similarity between adjacent vertebrae and the morphological variability of vertebrae. Most existing methods rely on fully supervised deep learning and, constrained by limited annotations, struggle to remain robust in complex scenarios. Methods: We propose a semi-supervised approach built on a dual-branch 3D U-Net. Mamba modules are inserted between the encoder and decoder to model long-range dependencies along the cranio–caudal axis. The identification branch employs a 3D convolutional block attention module (3D-CBAM) to enhance class discriminability. A unified semi-supervised objective is formulated via teacher–student consistency: for each unlabeled sample, weakly and strongly augmented views are generated, and cross-branch consistency is enforced, together with confidence-based filtering and class-frequency reweighting. In addition, a connected-component analysis is used to enforce anatomically plausible sequential continuity of vertebral indices in the outputs. Results: Experiments on VerSe 2019 and 2020 show that, on the public VerSe 2019 test set (with VerSe 2020 scans used as unlabeled training data), the supervised baseline achieved a Dice score of 89.8% and an identification accuracy of 92.3%. Incorporating unlabeled data improved performance to 91.6% Dice and 97.5% identification accuracy (relative gains of +1.8 and +5.2 percentage points). Compared with competing methods, the proposed semi-supervised model attains higher or comparable segmentation accuracy and the highest identification accuracy. Conclusions: Without additional annotation cost, the proposed method markedly improves the overall performance of vertebra segmentation and identification, offering more robust automated support for clinical workflows. Full article

Rice seed vigor is one of the critical factors determining rice yield and quality. Identifying substances related to seed vigor and rapidly assessing seed vigor by non-destructive methods are of great significance for increasing rice production. This study employed near-infrared diffuse reflectance spectroscopy (NIR-DRS) and transmission spectroscopy (NIR-TS) to evaluate the vigor of naturally aged rice seeds. The NIR-DRS failed to establish a reliable relationship between spectral data and seed vigor, proving ineffective in distinguishing seed vigor. After enhancing the spectral differences between viable and non-viable seeds, the NIR-TS successfully identified high-vigor and non-viable seeds, with a partial least squares discriminant analysis (PLS-DA) model achieving accuracy and germination rates of 84.52% and 88.57% on the test set, respectively. Furthermore, three algorithms, including interval partial least squares (iPLS), genetic algorithm (GA), and competitive adaptive reweighted sampling (CARS), were applied to extract characteristic spectral wavelengths associated with seed vigor. Among these, the CARS algorithm performed the best, identifying 38 characteristic wavelengths. Wavelength analysis indicated that rice seed vigor is primarily influenced by molecules such as starch, protein, moisture, and lipids. Using the characteristic wavelengths selected by the CARS algorithm, a PLS-DA prediction model for rice seed vigor was constructed, achieving high accuracy and germination rates of 90.47% and 95.38% on the test set, respectively. This study demonstrates that NIR-TS outperforms NIR-DRS in assessing rice seed vigor. Moreover, wavelength selection techniques can effectively identify characteristic spectral features related to seed vigor and significantly enhance the prediction accuracy of the model. Full article

Deployed in distributed organizations and edge networks, contemporary intrusion detection increasingly requires high-performing models without centralizing sensitive traffic logs. This study presents a lightweight federated intrusion detection framework that integrates (i) unidirectional serialization to convert tabular flow records into short sequences, (ii) multi-scale one-dimensional convolutions to capture heterogeneous temporal–statistical patterns at different receptive fields, and (iii) an attention-based reweighting module that emphasizes informative feature channels prior to classification. A sample-size-weighted FedAvg aggregation protocol is used to train a global detector without transferring raw data. Experiments on three widely used benchmarks (UNSW-NB15, KDD Cup 99, and NSL-KDD) under multiple client configurations report consistently high detection effectiveness, with peak accuracies of 99.38% (UNSW-NB15), 99.86% (KDD Cup 99), and 99.02% (NSL-KDD), alongside strong precision, recall, and F1 scores. In addition, the proposed framework is quantitatively benchmarked on UNSW-NB15 against two recent federated intrusion detection baselines, FedMSP-SPEC and a multi-view federated CAE-NSVM model, demonstrating improvements of more than 10 percentage points in macro F1-score while retaining a compact architecture. The manuscript further specifies a concrete threat model, clarifies the client data partitioning strategy and Non-IID quantification, and provides a reproducibility protocol (hyperparameters, random seeds, and evaluation procedures) to facilitate independent verification. Full article

The chemical composition of vegetable cooking oils is a key parameter in determining the quality of their products. Antioxidants are widely used in these products to extend their shelf life. In this study, the concentration of butylated hydroxytoluene (BHT) in edible oil was quantitatively determined by Raman spectroscopy combined with chemometrics. Initially, Raman spectra of edible oil samples with varying concentrations of BHT were obtained. Subsequently, three variable selection methods were applied to the pre-processed spectra. Optimised characteristic wavelengths were then used to establish a Radial Basis Function (RBF) neural network and partial least squares (PLS) models. The impact of variable selection on feature wavelengths was evaluated for both models in both independent and combined cases. The results demonstrate that the features identified through multiple variable selection methods correlate highly with the BHT content and can be utilised to develop high-precision detection models. The findings indicate that the PLS model, optimised using competitive adaptive reweighting (CARS), achieved the best prediction performance, with an average $R_P^2$ of 0.9687, and RMSEP of 3.1211. These results demonstrate the feasibility of using Raman spectroscopy combined with chemometrics for the rapid screening of BHT in edible oils. While the current study focuses on a broad concentration range to validate the method’s linearity, further optimisation is required for trace-level detection to meet strict regulatory limits. Full article

Connected and Autonomous Vehicles (CAVs) are exposed to increasingly sophisticated cyber threats hidden within high-dimensional, heterogeneous network traffic. A critical bottleneck in existing Intrusion Detection Systems (IDS) is the feature heterogeneity gap: discrete protocol signatures (e.g., flags, services) and continuous traffic statistics (e.g., flow duration, packet rates) reside in disjoint latent spaces. Traditional deep learning approaches typically rely on naive feature concatenation, which fails to capture the intricate, non-linear semantic dependencies between these modalities, leading to suboptimal performance on long-tail, minority attack classes. This paper proposes HCA-IDS, a novel framework centered on Semantics-Aware Cross-Modal Alignment. Unlike heavy-weight models, HCA-IDS adopts a streamlined Multi-Layer Perceptron (MLP) backbone optimized for edge deployment. We introduce a dedicated Multi-Head Cross-Attention mechanism that explicitly utilizes static “Pattern” features to dynamically query and re-weight relevant dynamic “State” behaviors. This architecture forces the model to learn a unified semantic manifold where protocol anomalies are automatically aligned with their corresponding statistical footprints. Empirical assessments on the NSL-KDD and CICIDS2018 datasets, validated through rigorous 5-Fold Cross-Validation, substantiate the robustness of this approach. The model achieves a Macro-F1 score of over 94% on 7 consolidated attack categories, exhibiting exceptional sensitivity to minority attacks (e.g., Web Attacks and Infiltration). Crucially, HCA-IDS is ultra-lightweight, with a model size of approximately 1.00 MB and an inference latency of 0.0037 ms per sample. These results confirm that explicit semantic alignment combined with a lightweight architecture is key to robust, real-time intrusion detection in resource-constrained CAVs. Full article

In autonomous driving systems, fog decreases visibility and contrast, blurs the boundary of objects, and increases scale variations, which results in missed detections. To solve these problems, we introduce Fog-YOLO, a lightweight real-time detector based on the YOLOv12 framework. First, we design the A2C2f-FSA module, which enhances the representation of fog-affected areas and low-contrast objects by modeling long-range dependencies in the frequency domain. This module effectively suppresses the interference of fog and background noise while maintaining low computational overhead. Second, we propose a bidirectional feature fusion module (BFFM) that uses decoupled attention paths to fuse deep semantic features and shallow texture details. This approach enhances robustness across multiple scales, ensuring the capture of fine-grained texture information and the preservation of global contextual information in foggy environments. Third, we introduce GSConv, which reduces parameters and computational cost by balancing spatial correlation modeling and computational complexity, optimizing the feature extraction process. Finally, we design the F-WIoU v3 loss function, which optimizes bounding box regression through dynamic focusing and difficulty re-weighting strategies, thereby reducing the influence of low-quality samples while improving the model’s localization robustness in foggy conditions. Experiments on the RTTS real-world fog dataset and the VOC-FOG synthetic dataset show that Fog-YOLO outperforms the baseline by 5.2% and 7.3% in mAP@0.5 with real-time inference speed. Overall, Fog-YOLO outperforms mainstream lightweight detectors, demonstrating its practical usefulness for autonomous driving in foggy environments. Full article