Search Results (10,525)

To address the limitation of insufficient mechanical strength and short service life in biodegradable bamboo fiber mulch film (BFM) replacing plastic film in agriculture, this study applied a biochemical method to make bamboo fiber and used bacterial cellulose (BC) as a natural nanoscale reinforcing agent to fabricate high-performance bacterial cellulose bamboo fiber mulch film (BC-BFM). The physical and mechanical properties, chemical structure, seed germination and degradation behavior performance of BC-BFM were characterized. Results demonstrated the structural compactness and homogeneity of the BC-BFM were improved markedly with the increase in BC addition and BC formed a 3D nanofibrillar network that effectively bridged inter-fiber voids. The tensile, burst and tear indexes of BC-BFM all significantly rose with BC addition. Notably, compared to plastic film and BFM, BC-BFM exhibited a good effect on mung bean seed germination and the best growth speed was at 5% BC addition. Furthermore, the degradation test showed that the degradation rate of BC-BFM within 90 d was three times less than that of BFM and service life was similar to plastic film. This showed that it was a promising method to prepare biodegradable high-quality BFM through biochemical preparation of bamboo fiber and BC nanocellulose reinforcement. This method markedly enhanced the mechanical performance and durability of BC-BFM, providing a feasible technical path for the development of biodegradable high-performance green agricultural covering materials with long service life. Full article

Baihua Reservoir, a typical large waterbody in the karst region of southwestern China and an essential drinking water source, is characterized by a high carbonate buffering capacity that profoundly shapes the structure and function of its phytoplankton community. This study systematically elucidates the multi-dimensional driving mechanisms underlying seasonal phytoplankton community assembly in karst reservoirs by integrating multiple analytical models—including the Neutral Community Model, β-diversity decomposition, co-occurrence network analysis, XGBoost-SHAP machine learning, and Partial Least Squares Path Modeling—based on monthly sampling at five sites from 2020 to 2024. The results revealed that: (1) Stochastic processes dominated community assembly across all four seasons, while deterministic processes played a crucial role in local species turnover. (2) The co-occurrence network structure showed significant seasonal dynamics, with the composition of keystone species adaptively shifting in response to changing environmental conditions. (3) The key environmental factors influencing the phytoplankton community exhibited clear seasonal patterns, primarily pH, NH₃-N, and COD_Mn in spring; water temperature, COD_Mn, and NH₃-N in summer; TN, TP, and pH in autumn; and pH, water temperature, and DO in winter. To support the sustainable management of karst reservoirs, we propose seasonally differentiated strategies derived from our phytoplankton community analysis: target COD_Mn reduction in spring and summer, focus on TN and TP load control in autumn, prioritize water column stability in winter, and maintain hydrological connectivity and pH monitoring year-round. This approach enhances phytoplankton community stability, safeguards drinking water safety, and provides a targeted management model for similar reservoir ecosystems globally. Full article

Robust plant disease detection in real-world agricultural environments remains challenging due to dynamic environmental conditions. Accurate and reliable disease identification is essential for precision agriculture and effective crop management. Although computer vision and Artificial Intelligence (AI) have shown promising results in controlled settings, their performance often drops under lesion scale variability, inter- and intra-class similarity among diseases, class imbalance, and illumination fluctuations. To overcome these challenges, we propose a Heterogeneous Feature Aggregation Network (HFA-Net) that brings together architectural improvements, illumination-aware preprocessing, and training-level enhancements into a single cohesive framework. To extract richer and more discriminative features from the early layers of the network, HFA-Net introduces a multi-scale, multi-level feature aggregation stem. The Reduction-Expansion (RE) mechanism helps preserve important lesion details while adapting to variations in scale. Considering real agricultural environments, an Illumination-Adaptive Contrast Enhancement (IACE) preprocessing pipeline is designed to address illumination variability in real agricultural environments. Experimental results show that HFA-Net achieves 96.03% accuracy under normal conditions and maintains strong performance under challenging lighting scenarios, achieving 92.95% and 93.07% accuracy in extremely dark and bright environments, respectively. Furthermore, quantitative explainability analysis using perturbation-based metrics demonstrates that the model’s predictions are not only accurate but also faithful to disease-relevant regions. Finally, Grad-CAM-based visual explanations confirm that the model’s predictions are driven by disease-specific regions, enhancing interpretability and practical reliability. Full article

The increasing demand for scalable and fair assessment of open-form responses in digital education shows the need for intelligent grading systems capable of balancing syntactic precision with semantic understanding. This study proposes a hybrid semantic–syntactic NLP framework for automated grading of short-answer and cloze-type questions. The framework integrates a rule-based matcher for syntactic accuracy, MPNet (Masked and Permuted Pre-trained Network) embeddings for semantic similarity, and a fine-tuned DeBERTa (Decoding-enhanced Bidirectional Encoder Representations from Transformer with Disentangled Attention) regressor for continuous score prediction, while a T5-small model provides pedagogically aligned feedback generation. Evaluations were conducted using benchmark corpora, synthetic cloze datasets, and a domain-specific short-answer corpus. Results demonstrate that the hybrid system outperforms traditional baselines, achieving 91% accuracy, a 0.89 F1 score, a mean absolute error of 0.36, and strong inter-rater agreement (κ = 0.87), aligning closely with human graders. Qualitative analyses show that the framework successfully recognizes paraphrased responses, assigns partial credit, and generates meaningful feedback. Ablation studies further validate the necessity of each subsystem, with performance significantly declining when components were removed. The findings confirm that the proposed framework is both computationally robust and pedagogically valuable, establishing a foundation for scalable, interpretable, and fair automated grading in contemporary educational environments. Full article

Fine-grained image classification under few-shot learning conditions remains a significant challenge due to limited labeled data and high intra-class similarity. This paper proposes a novel cross-modal framework that integrates Contrastive Language-Image Pretraining (CLIP) embeddings within a Siamese similarity network to enable robust and label-efficient classification. By leveraging the semantic alignment between textual class descriptions and visual representations, the model forms hybrid similarity pairs of image-to-image and image-to-text within a shared latent space, facilitating discriminative learning under low-shot scenarios. The architecture employs a dual-branch CLIP encoder and a contrastive loss function to optimize intra-class compactness and inter-class separability. Experiments conducted on benchmark datasets including miniImageNet and CUB-200-2011 demonstrate substantial improvements over zero-shot and few-shot baselines, achieving 70.32% accuracy, 71.15% F1-score, and 68.47% mAP on 5-way 1-shot and 78.41% accuracy, 79.02% F1-score, and 76.83% mAP on 5-way 5-shot tasks (averaged over 600 episodes with 95% confidence intervals on the CUB-200-2011 dataset). Extended evaluations under 10-way settings show similarly strong performance. Ablation studies further validate the critical roles of contrastive learning, normalization, and cross-modal embeddings in enhancing generalization. This work presents a scalable and interpretable paradigm for fine-grained classification in data-scarce domains. Full article

Accurate soil moisture assessment is essential for effective agricultural management in the southern US, where water availability has a significant impact on crop productivity. This study evaluates the Soil Moisture Active Passive (SMAP) Level-4 daily soil moisture product using in situ measurements from Natural Resources Conservation Service (NRCS) Soil Climate Analysis Network (SCAN) stations and the US. Climate Reference Network (USCRN) across diverse agroecosystems in Texas from 2016 to 2024. SMAP’s performance was examined across ten climate zones and six major land cover types, including urban regions, pastureland, grassland, rangeland, shrubland, and deciduous forests. Statistical metrics, including the coefficient of determination (R²), Root Mean Square Error (RMSE), Bias, and unbiased RMSE (ubRMSE) were used to evaluate the agreement between SMAP-derived and in situ soil moisture measurements. Results show that SMAP effectively captures seasonal soil moisture dynamics but exhibits spatially variable accuracy. The highest agreement was observed at Panther Junction (R² = 0.57, RMSE = 2.29%), followed by Austin (R² = 0.57, RMSE = 9.95%). While a weaker coefficient of determination was observed at PVAMU (R² = 0.28, RMSE = 11.28%) and Kingsville (R² = 0.11, RMSE = 7.33%), likely due to heterogeneity in land cover, and urbanized landscapes in these stations. Applying the quantile mapping bias correction methods significantly reduced RMSE and improved the accuracy of SMAP soil moisture data at some in situ measurement stations. The results highlight the importance of station-specific calibration and the integration of satellite and ground-based measurements to improve soil moisture monitoring for agriculture and drought management in Texas and similar regions. Full article

Accurate remote sensing scene classification is essential for large-scale Earth observation but remains challenging due to significant inter-class similarity and complex spatial layouts in medium- and low-resolution imagery. Conventional convolutional neural networks (CNNs) effectively capture local structural patterns but struggle to model long-range semantic dependencies, whereas Vision Transformers excel at global context modeling yet often show reduced sensitivity to fine-grained spatial structures. To address these limitations, we propose CrtNet, a structure-aware Cross-Model Residual Transformer Network that establishes a dual-stream collaborative architecture integrating convolutional structural representations with Transformer-based semantic modeling through gated residual cross-model interactions. In this framework, a convolutional branch first extracts stable local structural features with strong spatial inductive biases. These features are continuously injected into the Transformer encoding process via residual cross-model connections, enabling persistent structural guidance during global attention modeling. In addition, a sample-adaptive dynamic gating mechanism is introduced to flexibly balance structural and semantic features during prediction. Extensive experiments conducted on two public remote sensing benchmarks, EuroSAT and UCM, demonstrate that CrtNet consistently outperforms representative CNN-based, Transformer-based, and hybrid state-of-the-art models, particularly in visually ambiguous scene categories. Full article

The Vietnam Mekong River Delta (VMRD) is a climate-sensitive region characterized by diverse ecosystems, including extensive mangrove forests that protect against sea-level rise and contribute to global carbon sequestration. Accurate land cover classification in the VMRD is essential but remains challenging due to complex landscapes and dynamic environmental conditions. The primary objective of this study is to propose a semi-supervised deep learning framework that integrates satellite indices with multi-temporal remote sensing data to address key classification challenges, particularly in situations where ground truth data is limited, as compared to unsupervised and supervised machine learning methods. Our comparative analysis across different sample sizes (500 to 6000 ground-truth data points) reveals critical insights into model performance and scalability. Supervised models, including Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN), demonstrated strong performance when sufficient labeled data were available, with CNN achieving the highest accuracy (0.97 at 6000 samples). However, at minimal sample sizes (500 sample points), these supervised approaches exhibited substantial limitations, with accuracies dropping dramatically (RF: 0.75, SVM: 0.80, CNN: 0.81). Supervised models also showed overfitting tendencies compared to official land cover statistics. In contrast, the semi-supervised approach (SoC4SS-FGVC) achieves remarkably high performance at small sample sizes (0.92 accuracy with 500 sample points), demonstrating strength under minimal data availability. The framework also showed improved capability in distinguishing spectrally similar land-cover classes and detecting environmentally sensitive types such as mangrove forests. Cross-validation with official statistics confirmed the semi-supervised model’s superior effectiveness in delineating paddy rice fields and its resistance to overfitting. The performance analysis demonstrates that SoC4SS-FGVC provides a practical and cost-effective solution for land cover mapping, particularly in regions where extensive ground-truth data collection is prohibitively expensive or logistically challenging. Full article

Fluid mechanics in disordered structures gives rise to rich multiscale dynamics through the interplay of topology, symmetry breaking, and fluid–structure interactions. Heterogeneous networks encode mechanical responses, regulate flow organization, and shape energy dissipation, enabling memory effects and emergent collective behaviors across both natural and engineered systems. These principles operate across vast scales: from seamounts with characteristic scales of $L\approx {10}^3\mathrm{m}$ and Froude numbers $Fr\approx {10}^{-2}–{10}^{-1}$ generating deep-ocean turbulent mixing, to marine tidal turbines operating at Reynolds numbers $Re\approx {10}^7–{10}^8$ and Euler numbers $Eu\approx {10}^{-1}–{10}^0$, where inertial forces dominate flow dynamics. Although the dominant physical forces may vary across scales—for example, planetary rotation and stratification in large-scale oceanic flows versus viscous or interfacial effects in microscale systems—the comparison of dimensionless parameters provides a useful framework for discussing similarities in flow organization and scaling behavior. Empirical observations, network-based descriptions, and multiscale simulations collectively demonstrate how topological features constrain symmetry, organize transport pathways, and support predictive reconstruction and inverse design. These principles underpin applications ranging from engineered systems that exploit broken symmetries to rectify chaotic transport, to biological architectures where flows mediate information transfer, locomotion, and structural self-organization. In this Review, we synthesize recent advances to propose a unifying physical paradigm: fluid flows actively interact with disorder, reorganize dissipation, and convert structural asymmetry into functional mechanical performance across scales. Full article

(1) Background: The scarcity and imbalance of real fault data significantly limit the development of data-driven fault diagnosis methods for liquid rocket engines (LREs), especially under few-shot conditions. (2) Methods: To address this issue, this study proposes a GAN-based fault data augmentation framework for multivariate LRE time-series signals and a hybrid diagnostic classifier combining convolutional neural networks (CNNs), bidirectional long short-term memory (BiLSTM), and multi-head attention (MHA). The GAN component is introduced to alleviate fault-data scarcity and class imbalance by generating additional fault-like samples, while the classifier is designed to capture local features, long-range temporal dependencies, and diagnostically informative temporal regions. (3) Results: A multidimensional evaluation based on temporal similarity, statistical consistency, and global distribution discrepancy indicates that the generated samples preserve important characteristics of the original signals under the current evaluation protocol. On the augmented LRE dataset, the proposed classifier achieved strong diagnostic performance. In addition, supplementary experiments on the public HIT aero-engine dataset further support the effectiveness of the classifier architecture, its component-wise contribution, and its behavior under imbalanced few-shot settings, while also demonstrating the value of uncertainty-aware prediction. (4) Conclusions: The results provide encouraging evidence that the proposed framework can improve LRE fault diagnosis under data-scarce conditions. However, the present findings should be interpreted within the scope of the available data and evaluation setting. More comprehensive generator-side ablation, broader external validation, and physics-oriented assessment of the generated signals are still needed before stronger conclusions can be made. Full article

Introduction: Diabetic gastroparesis increases the risk of aspiration, pneumonia, and sepsis, yet the impact of glucagon-like peptide-1 receptor agonists (GLP-1 RAs) on these outcomes is uncertain because of their gastric-emptying effects. Methods: We performed a retrospective cohort study using the TriNetX Global Research Network. Adults (≥18 years) with diabetes mellitus and gastroparesis were identified and divided into two cohorts based on GLP-1 RA exposure. Propensity score matching (1:1) balanced demographics, comorbidities, and antidiabetic medications, yielding 23,371 patients per cohort. Outcomes, assessed from 180 days after index, included pneumonia, pneumonitis, mechanical ventilation, ventilator-associated pneumonia, sepsis, bacteremia, empyema, lung abscess, acute respiratory distress syndrome (ARDS), and need for enteral feeding. Risk ratios (RRs) and hazard ratios (HRs) with 95% confidence intervals (CIs) were estimated. Results: Compared with GLP-1 users, non-GLP-1 patients had higher incidences of pneumonitis (3.6% vs. 2.5%; HR 1.76, 95% CI 1.58–1.95), pneumonia (13.2% vs. 12.2%; HR 1.34, 95% CI 1.27–1.41), mechanical ventilation (4.4% vs. 3.3%; HR 1.63, 95% CI 1.49–1.79), sepsis (12.8% vs. 11.1%; HR 1.44, 95% CI 1.37–1.52), and bacteremia (5.2% vs. 4.4%; HR 1.46, 95% CI 1.35–1.59) (all p < 0.001). Empyema and ARDS were also numerically lower among GLP-1 users, while ventilator-associated pneumonia and lung abscess were rare and similar between groups. No patients required percutaneous endoscopic gastrostomy or nasal enteral feeding. Conclusions: In patients with diabetes and gastroparesis, GLP-1 RA therapy was associated with significantly fewer pulmonary and systemic infectious complications. These data suggest that the systemic benefits of GLP-1 RAs may outweigh concerns regarding delayed gastric emptying in this high-risk population. Full article

The comprehension of the complex dynamics of degassing is critical for volcano monitoring and assessing volcanic hazards. In this study, we apply visibility graph analysis (VGA) to a decadal, high-resolution time series of daily soil CO₂ flux recorded by a standardized monitoring network at Mt. Etna volcano (Italy). By mapping these time series into complex networks, we demonstrate that the connectivity degree distributions follow a power law described by the exponent $\gamma$, which reveals a self-similar behavior of gas emissions. We introduce the $\gamma$-deviation, namely the variation of the scaling exponent from its long-term site-specific baseline, as a novel proxy for degassing efficiency. The long-term baseline is interpreted as a site-specific measure of flux efficiency, while its variations are attributed to other factors, such as fluctuations in the sources or changes in the efficiency of fluids transport pathways. Our results identify a transition from a period of discordance across the monitoring sites (pre-2016) to a phase of network-wide concordance (after 2016). The striking correlation between topological $\gamma$-deviations and the established normalized network signal (${\Phi }_{norm}$) validates the methodology, suggesting that VGA is able to capture the same underlying magmatic drivers. This study establishes VGA as a robust and reliable tool for medium- and long-term monitoring, potentially capable of identifying the occurrence of large-scale magmatic processes and refining the characterization of fluid transport dynamics in active volcanic systems. Full article

Continuous representations in brain navigation system are manifested as spatially structured patterns of population activity, such as a single-peaked bump moving along a ring manifold in head-direction system and hexagonal lattice patterns underlying spatial representation in grid-cell systems. These phenomena are commonly modelled within the framework of continuous attractor networks (neural dynamical field), yet the mechanisms by which activation-function nonlinearities interact with connectivity structure to determine pattern selection and dynamics remain incompletely understood. This paper separately analyses the interactions between non-resonant and resonant modes using a multiscale unfolding approach. We show that, when the critical modes satisfy a resonance condition, the quadratic nonlinearity of the activation function induces a three-mode coupling that fundamentally alters the structure of the amplitude equations and becomes the dominant mechanism governing spatial pattern selection. Building on this analysis, we introduce a weak asymmetric component in the connectivity and analytically derive the resulting pattern drift velocity, which is subsequently confirmed by numerical simulations. Finally, we apply these dynamical mechanisms to input-driven scenarios, illustrating that similar dynamical mechanisms can account for activity-bump tracking in head-direction models and lattice translations in grid-cell models. Overall, this work provides an analytically tractable framework for studying pattern dynamics in neural field models relevant to spatial representations, and may inform biomimetic approaches to spatial representation and navigation. Full article

Authenticity verification for thangka artworks remains challenging in the market, as traditional physical authentication methods offer limited reliability, while modern spectroscopic and chemical testing technologies are costly and unsuitable for large-scale application. Although deep learning methods can achieve efficient authentication through image features, they rely on centralized databases to store feature information, making them susceptible to tampering risks and undermining the credibility of authentication results. To address these issues, this study proposes a digital authentication method for thangka paintings that integrates blockchain technology. This approach stores image features in the InterPlanetary File System (IPFS) and records their hash values on the blockchain, ensuring the immutability and verifiable evidence of feature data. Simultaneously, it employs convolutional neural networks for feature extraction and similarity analysis of thangka images, constructing an integrated platform system encompassing storage, authentication, and traceability. This enhances the reliability and automation of authentication outcomes. The platform further supports full-process traceability of thangka storage and authentication operations, providing a viable pathway for establishing a scientific and reliable digital authentication system for thangkas. Experimental evaluation on a dataset of 2847 thangka images demonstrates 99.2% authentication accuracy, with a precision of 98.7% and an F1-score of 99.1%, while end-to-end authentication latency averages 1247 ms, validating the system’s effectiveness for practical museum and market deployment scenarios. Full article

The threat of plant diseases in economically significant crops of the Solanaceae family, especially tomatoes and potatoes, is a significant challenge to global food security, highlighting the necessity of fast and convenient diagnostic methods. This paper introduces an enhanced MobileNetV2 model to perform automated disease classification through the use of a domain-specific self-supervised learning (SSL) pretraining approach. The model was first trained on 54,303 unlabeled plant images to learn basic botanical representations, followed by fine-tuning under six experimental conditions to optimize disease classification performance. Findings show that SSL pretrained weights consistently outperform traditional ImageNet-based transfer learning, achieving 0.9158 overall accuracy and a weighted F1-score of 0.9143 in joint tomato and potato classification. The model demonstrates strong cross-crop generalization, correctly identifying Early Blight and Late Blight with accuracies of 0.9600 and 0.9359, respectively, and effectively separating disease-specific visual symptoms from host morphology. Confusion matrix analysis further indicates a reduction in misclassification of visually similar necrotic lesions, a common challenge in supervised models. Overall, the proposed SSL architecture enhances the performance of lightweight convolutional neural networks (CNNs) to a large extent, providing a strong, computationally efficient solution for field-deployable diagnostics in precision agriculture, particularly for tomato and potato crops. Full article