Search Results (561)

Text-attributed graphs (TAGs) couple graph topology with node-level text, but real data often contain spurious edges, missing links, and text–structure mismatch that destabilize learning under scarce labels. We propose CoACL (Coupled Augmentation for Contrastive Learning), a framework that uses LLM semantic supervision to denoise structural and textual information and alleviate data sparsity. CoACL first prunes the candidate edge space using structural similarity and then queries an LLM to discard suspicious edges and confirm plausible links, yielding semantically consistent positive and negative pairs. We further introduce keyword-focused text augmentations and learn coupled representations by optimizing a joint text–graph contrastive objective guided by semantics. Experiment results on Cora, PubMed, and the Open Graph Benchmark Arxiv dataset (OGBN-Arxiv) show that CoACL consistently outperforms strong baselines and yields up to 7.1% absolute improvement in node classification accuracy, with the largest gains in low-label regimes. By constraining LLM evaluation to similarity-based candidates, CoACL targets neighborhood-level noise with controlled cost. Full article

Visual exploration of time series data is essential for uncovering meaningful insights in domains such as healthcare monitoring and financial analysis, yet it remains computationally challenging due to the combinatorial explosion of potential subsequence comparisons. For long time series, an exhaustive comparison of all possible subsequence pairs becomes prohibitively expensive, limiting interactive exploration. This paper presents the TiVEx (Time Series Visual Exploration) family of algorithms for efficiently discovering the top-k most dissimilar subsequence pairs in comparative time series analysis. TiVEx achieves scalability through three complementary strategies: TiVEx-sharing exploits computational reuse across overlapping subsequence windows, eliminating redundant distance calculations; TiVEx-pruning employs distance-based upper bounds to eliminate unpromising candidates without exhaustive evaluation; and TiVEx-hybrid integrates both mechanisms to maximize efficiency gains. The key observation is that overlapping subsequences share a substantial computational structure, which can be systematically exploited while maintaining result optimality through provably correct pruning bounds. Extensive experiments on six diverse datasets demonstrate that TiVEx-hybrid achieves up to 84% reduction in distance calculations compared to exhaustive search while producing identical top-k results. Compared to state-of-the-art subsequence comparison methods, TiVEx-hybrid achieves 2.3× improvement in computational efficiency. Our effectiveness analysis confirms that TiVEx achieves result quality within 5% of exhaustive search even when exploring only a subset of candidate positions, enabling scalable visual exploration without compromising insight quality. Full article

This study aimed to evaluate the aqueous extraction of phenolic compounds from olive tree pruning. Soxhlet extraction and aqueous extraction at 120 °C were performed in two types of pressurized reactors and different scales. The highest total phenolic content was obtained using Soxhlet (3809.8 mg/100 g biomass), followed by the other extraction strategies (up to 1500 mg/100 g). The content of 3,4-dihydroxyphenylglycol, hydroxytyrosol, and oleuropein also varied depending on the extraction conditions. Overall, aqueous extraction at 120 °C can be used to partially recover phenolic compounds, albeit in a shorter time compared to Soxhlet extraction and using higher solid loads to facilitate scaling up. This type of extraction can be applied in the future to recover these high-value compounds from olive tree pruning, a common agricultural byproduct of the Mediterranean region. Full article

Filter pruning is an effective approach for improving the inference efficiency of neural networks and is particularly attractive for on-device artificial intelligence (AI) applications. However, many existing methods fail to accurately identify redundant filters due to limited modeling of inter-filter dependencies. A filter pruning method based on nuclear norm analysis is proposed to quantify filter independence and guide structured pruning. By analyzing the layer-wise distribution of independence scores, a principled trade-off between pruning rate and accuracy preservation is achieved. In most evaluation scenarios, the proposed method achieves 75–95% parameter reduction and 70–80% FLOPs reduction, while substantially higher compression ratios (up to 99%) can be obtained for more redundant network architectures, with consistent performance trends observed across multiple accuracy-related metrics. Furthermore, deployment on an RK3588 neural processing unit (NPU) demonstrates substantial reductions in memory consumption and inference latency, confirming the practical effectiveness of the method for mobile and edge AI applications. Full article

The integration of Federated Learning (FL) and Graph Neural Networks (GNNs) has emerged as a promising paradigm for distributed intelligence in Internet of Things (IoT) environments. However, challenges related to trust, device heterogeneity, and energy efficiency continue to hinder scalable deployment in real-world settings. This paper presents Trust-FedGNN, a trust-aware federated graph learning framework that jointly addresses reliability, robustness, and sustainability in IoT ecosystems. The framework combines reliability-based reputation modeling, energy-aware client scheduling, and dynamic graph pruning to reduce communication overhead and energy consumption during collaborative training, while mitigating the influence of unreliable or malicious participants. Trust evaluation is explicitly decoupled from energy availability, ensuring that honest but resource-constrained devices are not penalized during aggregation. Experimental results on benchmark IoT datasets demonstrate up to 5.8% higher accuracy, 3.1% higher F1-score, and approximately 22% lower energy consumption compared with State-of-the-Art federated baselines, while maintaining robustness under partial adversarial participation. These results confirm the effectiveness of Trust-FedGNN as a secure, robust, and energy-efficient federated graph learning solution for heterogeneous IoT networks (a proof-of-concept evaluation across 10 federated clients). Full article

Recycling vineyard pruning residues into compost and vermicompost represents a sustainable strategy to reduce viticulture’s reliance on chemical fertilizers. Nonetheless, their effects on plant performance remain poorly understood. This study evaluated the effect of vine pruning residues compost and vermicompost on the physiological, biochemical, and growth performance of Vitis vinifera L. cv. Touriga Franca, in comparison with mineral fertilization and an unfertilized control. A pot experiment was conducted from April to September 2024 in northern Portugal under Mediterranean climate conditions, using one-year-old grapevines and subjected to four fertilization treatments. Leaf gas exchange, chlorophyll a fluorescence, photosynthetic pigments, antioxidant and osmoprotective metabolites, and shoot and root development were assessed at three sampling dates during the growing season. Organic amendments enhanced photosynthetic performance and root growth relative to the unfertilized control. Vermicompost promoted higher CO₂ assimilation, stomatal conductance, and shoot and root elongation, whereas compost increased intrinsic water use efficiency, photochemical regulation, and root biomass. Biochemical analyses indicated that compost favored protein and carotenoid accumulation, while vermicompost increased proline and later protein levels, alongside reduced phenolic and flavonoid contents. Despite their similar chemical composition, compost and vermicompost induced distinct physiological responses driven by differences in biological activity and nutrient dynamics. These findings demonstrate that pruning-derived organic amendments can match mineral fertilization in supporting grapevine performance while offering additional benefits for stress regulation and sustainable vineyard management. Full article

This study addresses the challenge of instance segmentation for key grapevine structures (Trunk, Branch, and Bud) in complex natural environments, focusing on issues such as varying light conditions, weather, and significant scale variations. We propose an enhanced instance segmentation model named CDA-YOLOv8. Trained on a self-built dataset of 2160 images covering grapevine scenes under diverse lighting conditions, this model integrates three key components: the ACmix module for enhanced global feature modeling, the C2f-DWR module for optimized multi-scale feature extraction, and the CSPPC module for achieving model lightweighting. We evaluate performance using precision/recall and mAP@50, together with the stricter mAP@[50:95], for segmentation quality, and parameters/model size/FPS for deployment efficiency. Experimental results demonstrate that CDA-YOLOv8 achieves 70.1% precision, 74.4% recall, 76.3% mAP@50, and 36.8% mAP@[50:95], with only 3.19 million parameters and a compact model size of 6.49 MB. Compared with the original YOLOv8-seg, CDA-YOLOv8 improves segmentation accuracy while maintaining high efficiency (6.87 FPS). It also delivers better mask quality under stricter overlap criteria, providing quantitative evidence for real-time perception in automated grapevine pruning systems. Full article

Advances in medical care now enable significant functional recovery after traumatic limb injuries. The return-to-duty decision-making process is highly variable and dependent on multiple factors. To retain service members (SM) post-injury, there needs to be a robust method to inform the decision-making process. The collection of outcome data and decision tree analysis has the potential to assist in the development of an efficient decision support tool. Data were combined from two previous research studies on 31 injured SMs (26 with limb salvage wearing custom dynamic ankle–foot orthoses and 5 with varying levels of lower limb amputation wearing prostheses). Forty-two factors across military, demographic, injury, and outcome measures were used to develop categorical tree models to classify return to duty after injury. The feasibility of the final pruned model was evaluated using a 10-fold cross-validation to calculate sensitivity, specificity, and misclassification rate. The overall misclassification rate for the final pruned model was 29% (9/31). The model classified participants into successful return to duty: (1) Post Concussion Symptom Scale < 20 and (2) age at time of assessment ≥34. These preliminary results suggest that decision tree modeling could be an effective approach to augmenting the return-to-duty decision-making process. Full article

AI-driven network security relies increasingly on Large Language Models (LLMs) to detect sophisticated threats; however, their deployment on resource-constrained edge devices is severely hindered by immense parameter scales. While unstructured pruning offers a theoretical reduction in model size, commodity Graphics Processing Unit (GPU) architectures fail to efficiently leverage element-wise sparsity due to the mismatch between fine-grained pruning patterns and the coarse-grained parallelism of Tensor Cores, leading to latency bottlenecks that compromise real-time analysis of high-volume security telemetry. To bridge this gap, we propose SPARTA (Sparse Parallel Architecture for Real-Time Threat Analysis), an algorithm–architecture co-design framework. Specifically, we integrate a hardware-based address remapping interface to enable flexible row-offset access. This mechanism facilitates a novel graph-based column vector merging strategy that aligns sparse data with Tensor Core parallelism, complemented by a pipelined execution scheme to mask decoding latencies. Evaluations on Llama2-7B and Llama2-13B benchmarks demonstrate that SPARTA achieves an average speedup of 2.35× compared to Flash-LLM, with peak speedups reaching 5.05×. These findings indicate that hardware-aware microarchitectural adaptations can effectively mitigate the penalties of unstructured sparsity, providing a viable pathway for efficient deployment in resource-constrained edge security. Full article

This paper presents the FPGA implementation and verification of a lightweight one-dimensional convolutional neural network (1D-CNN) pipeline for real-time battery state-of-charge (SoC) estimation in automotive battery management systems. The proposed model employs separable 1D convolution and global average pooling, and applies aggressive structured pruning to reduce the number of parameters from 3121 to 358, representing an 88.5% reduction, without significant accuracy loss. Using quantization-aware training (QAT), the network is trained and executed in INT8, which reduces weight storage to one-quarter of the 32-bit baseline while maintaining high estimation accuracy with a Mean Absolute Error (MAE) of 0.0172. The hardware adopts a time-multiplexed single MAC architecture with FSM control, occupying 98,410 gates under a 28 nm process. Evaluations on an FPGA testbed with representative drive-cycle inputs show that the proposed INT8 pipeline achieves performance comparable to the floating-point reference with negligible precision drop, demonstrating its suitability for in-vehicle BMS deployment. Full article

Grayscale-based Encryption-then-Compression (EtC) systems transform RGB images into the YCbCr color space, concatenate the components into a single grayscale image, and apply block permutation, block rotation/flipping, and block-wise negative–positive inversion. Because this pipeline separates color components and disrupts inter-channel statistics, existing extended jigsaw puzzle solvers (JPSs) have been regarded as ineffective, and grayscale-based EtC systems have been considered resistant to ciphertext-only visual reconstruction. In this paper, we present a practical ciphertext-only attack against grayscale-based EtC. The proposed attack introduces three key components: (i) Texture-Based Component Classification (TBCC) to distinguish luminance (Y) and chrominance (Cb/Cr) blocks and focus reconstruction on structure-rich regions; (ii) Regularized Single-Channel Edge Compatibility (R-SCEC), which applies Tikhonov regularization to a single-channel variant of the Mahalanobis Gradient Compatibility (MGC) measure to alleviate covariance rank-deficiency while maintaining robustness under inversion and geometric transforms; and (iii) Adaptive Pruning based on the TBCC-reduced search space that skips redundant boundary matching computations to further improve reconstruction efficiency. Experiments show that, in settings where existing extended JPS solvers fail, our method can still recover visually recognizable semantic content, revealing a potential vulnerability in grayscale-based EtC and calling for a re-evaluation of its security. Full article

Objectives: To improve the differentiation of borderline ovarian-adnexal tumours (BOTs) from malignant ovarian-adnexal masses, most of which fall into the indeterminate O-RADS MRI 4 category, by developing a multimodal predictive model that integrates clinical, analytic, and MRI parameters. Methods: This retrospective, single-centre study included 248 women who underwent standardised MRI for ovarian-adnexal mass characterisation between 2019 and 2024. Of these, 201 had true ovarian-adnexal masses (114 benign, 22 borderline, and 65 malignant), confirmed by histopathology or stability after ≥12-month follow-up. Forty-one clinical, laboratory, and imaging variables were initially assessed, and after a bivariate evaluation, 18 final predictors with clinical relevance were selected for model construction with thresholds learned from the data. A classification and regression tree (CART) model (“Full Model”) was applied as a second-stage tool after O-RADS MRI scoring, using 10-fold cross-validation to prevent overfitting. A pruned “Simplified Model” was also derived to enhance interpretability. Results: O-RADS MRI performed well at the extremes (scores 2–3 and 5) but showed limited discrimination between BOTs and malignancies within category 4 (PPV for borderline = 0.50). The decision-tree models significantly improved diagnostic performance, increasing overall accuracy from 0.856 with O-RADS MRI alone to 0.905 (Simplified Model) and 0.955 (Full Model). The PPV for BOTs within the intermediate O-RADS MRI 4 category increased from 0.49 with O-RADS MRI alone to 0.77 and 0.90 with the simplified and full models, respectively, while maintaining high accuracy for benign and malignant lesions. Conclusions: In this retrospective single-centre cohort, the addition of an interpretable rule-based predictive model as a second-line tool within O-RADS MRI category 4 was associated with improved discrimination between borderline and invasive malignant ovarian-adnexal tumours. These findings suggest that multimodal integration of clinical, laboratory, and MRI features may help refine risk stratification in indeterminate cases; however, external validation in prospective multicentre cohorts is required before clinical implementation. Full article

Intensive agriculture has intensified soil acidification in southern China, threatening crop productivity and ecosystem sustainability. Biochar can neutralize acidity, improve pH buffering, and enhance nutrient retention and microbial habitat in acidic soils. Accordingly, we produced biochars from pruned eucalyptus (ABC), camphora (ZBC), and guava (FBC) branches via pyrolysis at 500 °C. The three biochars were characterized by elemental analysis, Fourier Transform Infrared Spectroscopy (FTIR), and SEM (Scanning Electron Microscopy), and their effects on soil properties, enzyme activities, and bacterial communities were evaluated through a 56-day incubation experiment in an acidified, continuously cropped soil. Physicochemical characterization revealed that ZBC and FBC possessed more oxygen-containing functional groups and greater potential for pH buffering and nutrient release, whereas ABC exhibited higher aromaticity and structural stability. Biochar significantly increased soil pH by 0.62–1.42 units and improved nutrient availability and carbon pools (p < 0.05). Additionally, 4% ZBC increased urease and sucrase activities by 21.54% and 79.34%, respectively, while 2% FBC increased cellulase activity by 25.99%. High-throughput sequencing identified Acidobacteria and Proteobacteria as the dominant phyla; ZBC and FBC at 0.5% and 2% significantly increased Shannon and Chao1 indices. Redundancy analysis indicated that available potassium, pH, soil organic carbon, urease, sucrase, and cellulase were the primary drivers of bacterial community variation and positively associated with carbon-cycling phyla. These findings demonstrate that feedstock-specific biochar properties critically regulate soil biogeochemical processes, offering a sustainable strategy to remediate acidified soils and valorize agroforestry residues. Full article

Fertilization management is crucial mainly during the walnut training phase in order to obtain good plant formation, which is essential for guaranteeing future optimal yield. The aim of the present experiment was to evaluate the effect of different organic amendments on plant nutritional status and soil fertility in young bearing walnut trees. The experiment was conducted in 2023 and 2024 on walnut trees of the cultivar Chandler grafted on Juglans regia, planted in 2021. Since 2023, plants were yearly treated as follows: 1. non-fertilized control; 2. mineral fertilization; 3. application of municipal solid waste compost; and 4. application of compost from agri-food chain scraps. Soil amendments were supplied at the same rate as mineral fertilizer (120 kg N ha⁻¹) in spring on the tree row on a 1.5 m wide strip, while mineral fertilizer was split in two applications (50% in spring and 50% in summer). Plant growth, measured with trunk diameter and pruning wood weight, was enhanced by mineral fertilization, followed by compost, in comparison to the control. Soil mineral N was too high in relation to plant needs, with a consequent increase in the risk of nitrate leaching. Organic amendments increased soil nutrient availability, microbial activity, and carbon concentration, which, in the long term, could provide a positive environmental effect related to its sequestration into the soil. Full article

The performance of convolutional neural networks (CNNs) depends not only on model architecture but also on the structure and quality of the training data. While most artificial network interpretability methods focus on explaining trained models, less attention has been given to understanding how dataset composition itself shapes learning outcomes. This work introduces a novel framework that uses programmatically generated synthetic datasets to isolate and control visual features, enabling systematic evaluation of their contribution to CNN performance. Guided by principles from set theory, Shapley values, and the Apriori algorithm, we formalize an equivalence between CNN kernel weights and pattern frequency counts, showing that feature overlap across datasets predicts model generalization. Methods include the construction of four synthetic digit datasets with controlled object and background features, training lightweight CNNs under K-fold validation, and statistical evaluation of cross-dataset performance. The results show that internal object patterns significantly improve accuracy and F1 scores compared to non-object background features, and that a dataset similarity prediction algorithm achieves near-perfect correlation ($\rho =0.97$) between the predicted and observed performance. The conclusions highlight that dataset feature composition can be treated as a measurable proxy for model behavior, offering a new path for dataset evaluation, pruning, and design optimization. This approach provides a principled framework for predicting CNN performance without requiring full-scale model training. Full article