Search Results (6,822)

This study proposes a new formalized approach to the stabilization of linear transformations in artificial neural networks, based on discrete algebraic properties. In contrast to existing stability methods that rely on spectral norms, regularization techniques, or empirical heuristics, this work introduces the concept of algebraic stabilization—stability that arises from the structural properties of the matrices defining linear operators. The central object of investigation is the class of integer-valued matrices for which exponentiation to a form of the type ${\mathrm{W}}^{\mathrm{k}}=\mathrm{I}+\mathsf{\mu }\mathrm{D}$ is possible, where $\mathrm{D}\in {\mathbb{Z}}^{\mathrm{n}\times \mathrm{n}},\mathsf{\mu }\in {\mathbb{Z}}_{>1}$. A well-known problem in group algebra is considered that guarantees the existence of such an exponent under the condition that $\mathsf{\mu }$ is coprime with the determinant of $\mathrm{W}$. Within this framework, modular arithmetic, reduction modulo $\mathsf{\mu }$, and the group structure of ${\mathrm{G}\mathrm{L}}_{\mathrm{n}}\left({\mathbb{Z}}_{\mathsf{\mu }}\right)$ are employed, thereby linking the proposed method to the theory of finite groups and linear automata. The advantages of the approach are discussed, including formal control over the iterative behavior of transformations, compatibility with quantized and finitely representable networks, the possibility of embedding stabilizing conditions directly into the network architecture, and the potential to improve model interpretability and reliability. At the same time, limitations are identified, particularly those related to constructive implementation, the selection of suitable hyperparameters, and generalization to broader classes of transformations. Full article

Age-related macular degeneration (AMD) is the leading cause of central vision loss in aging populations. Geographic atrophy (GA) is the advanced, non-neovascular form of AMD. Predicting the longitudinal progression of GA remains a critical challenge in ophthalmic clinical practice and clinical trial design. Forecasting the trajectory of GA is complicated by highly variable growth rates and the inherent scarcity of long-term, high-quality imaging data. To address these challenges, we introduce the Sliding Window Attention U-Net (SWAU-Net), a hybrid architecture that integrates Transformer-based temporal modeling of GA growth with precise spatial modeling of GA location with a U-Net convolutional neural network (CNN). To ensure generalization in the low-data regime, SWAU-Net embeds explicit temporal and geometric consistency priors via a weight-shared Sliding Window Attention core and feature-level regularization that preserves sparse, high-frequency lesion boundaries across frames. Experimental results demonstrate that these structural constraints prevent the model from overfitting to imaging noise, achieving a Growth Mask Dice Similarity Coefficient (DSC) of 0.66 (representing the spatial overlap between the predicted and ground truth lesion expansion regions), a significant improvement over unregularized Transformer and standard recurrent baseline models. Our framework provides a robust tool for predicting GA lesion trajectories, potentially supporting more efficient clinical trial designs and personalized patient monitoring. Full article

The increasing digital integration of Industrial Control Systems (ICS), including Supervisory Control and Data Acquisition (SCADA) and Distributed Control Systems (DCSs), has improved operational efficiency while simultaneously increasing exposure to cyber threats. Traditional signature-based intrusion detection systems are limited in detecting novel and stealthy attacks in dynamic industrial environments. This study presents a deep learning–based anomaly detection framework for ICS cybersecurity using multivariate time-series data from sensors, actuators, and network traffic. Three architectures, Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, and Transformer models, are evaluated using the HAI Security Dataset. Experimental results show that the Transformer model achieves the highest accuracy (92%), followed by CNN (91%) and LSTM (90%), with all models attaining an F1-score of 91%. The Transformer demonstrates superior generalization by effectively modelling complex temporal dependencies. Key challenges, including data imbalance, overfitting, and limited interpretability, are discussed alongside potential mitigation strategies such as hybrid modelling, federated learning, and digital twin integration. The findings demonstrate the effectiveness of deep learning for scalable, real-time cybersecurity threat detection in industrial control environments. To address challenges such as class imbalance and overfitting, the study discusses mitigation strategies including regularization, early stopping, cost-sensitive learning, and future integration of data balancing and federated learning techniques for improved robustness and generalization. Full article

In the context of accelerating worldwide urbanization and ecosystem decline, deciphering the interactions between landscape patterns and habitat quality is essential for biodiversity preservation, particularly within ecologically sensitive zones like the Yellow River Basin. This research investigates the spatiotemporal dynamics, spatial linkages, and nonlinear relationships connecting landscape patterns and habitat quality across the basin. Utilizing land use datasets spanning 1980–2023, we combined the InVEST model, landscape pattern indices, spatial autocorrelation analysis, the XGBoost algorithm, and SHAP interpretability methods. The results show that: (1) Landscape patterns underwent a clear transition around 1995, shifting from regularization and connectivity toward fragmentation and heterogeneity, evidenced by increases in PD, LSI, and SHEI, alongside decreases in LPI and CONTAG. (2) Mean habitat quality progressively declined, exhibited a spatial distribution characterized by “higher in the west, lower in the east.” Low-quality habitat areas expanded from 2.12% to 3.76%, whereas high-quality areas decreased from 23.12% to 22.45%, with better habitats largely maintained in western headwaters and the Qinling Mountains. (3) Significant spatial correlations were observed: LPI positively correlated with habitat quality, while PD, LSI, SHEI, and CONTAG showed negative correlations. Two dominant spatial aggregations emerged—namely “high connectivity–high quality” in the west and “high fragmentation–low quality” in the east. (4) CONTAG was identified as the dominant factor influencing habitat quality, with all landscape indices exhibiting distinct threshold effects. The proposed framework, which integrates spatial statistics, machine learning, and interpretability methods, offers a novel approach for deciphering complex ecological processes. Moreover, the identified thresholds and zonal management strategies offer a scientific foundation for ecological conservation and spatial planning in the Yellow River Basin and other vulnerable river systems worldwide. Full article

Rapid N-1 security assessment in modern power systems faces a critical conflict between computational timeliness and the heavy reliance on labeled data for high-fidelity models. To mitigate this issue, a unified framework co-optimizing a physics-informed neural network (PINN) and memory-resident computing is proposed. At the algorithm level, power flow equation residuals are incorporated into the PINN formulation as physical regularization terms. This integration facilitates better alignment with electrical constraints and improves generalization capabilities under small-sample conditions. At the system level, a heterogeneity-aware asynchronous parallel computing architecture is developed. In this architecture, pull-based scheduling and lock-free memory mapping are utilized to mitigate straggler effects, thereby reducing synchronization latency and I/O overhead. Numerical case studies on the IEEE 39-bus system demonstrate that the physics mismatch is reduced by nearly two orders of magnitude compared to a baseline deep neural network (DNN), and the total execution time for scanning 20,000 contingencies is decreased by 34.0%. Full article

Background/Objectives: In Phenylketonuria (PKU), engaging in regular physical activity and exercise (PA/E) is important for physical and psychological health, but additional considerations may be required to facilitate uptake and performance as well as to optimise metabolic control. The aim of this study, therefore, was to investigate the stakeholder perspectives on the barriers, facilitators, and solutions to completing PA/E, sport, and nutrition in PKU. Methods: In total, 7 in-person and 6 online semi-structured focus groups (FGs) were conducted with individuals with PKU (n = 31), caregivers (n = 13), clinicians (n = 17), and medical industry professionals (n = 14) in PKU (n = 75 total participants). Three main questions about the barriers, facilitators, and solutions to performing PA/E with PKU were explored. Identified themes were mapped onto the capability, opportunity, motivation, and behaviour (COM-B) model of behaviour change with anonymous quotes from relevant stakeholders used to illustrate the findings. Results: Five common themes were identified. Most notably, individuals with PKU and their caregivers stated fatigue, poor recovery, low energy, and fear around the impact of exercise on blood phenylalanine (Phe) control were barriers to PA/E. Individuals with PKU were aware of the potential benefits of exercise, stating PA/E impacted positively on their mental well-being, daily functioning, and happiness and improved their self-confidence and long-term health. Identified solutions to PA/E participation included greater knowledge in regard to the impact of PA/E on Phe levels, improvements in advice on amount and supplementation with protein substitutes, tailored PKU nutritional advice, more awareness of PA/E within and outside the PKU community, specific PKU guidelines for PA/E, more scientific research, and PA/E events. Misalignment was evident, such that individuals with PKU reported additional barriers to PA/E, whereas other key stakeholder groups perceived the same barriers as the general public. Conclusions: There seems to be a misalignment between individuals with PKU, caregivers, clinicians, and industry professionals regarding PA/E, sport, and nutrition. Individuals with PKU and caregivers reported additional barriers to undertaking PA/E, sport, and nutrition compared to the general public. This suggests that further education and collaboration is needed through stakeholders to better understand how such barriers could be overcome in respect of PA/E, sport, and nutrition in individuals with PKU. Full article

Addressing the challenges in vision-based sensor networks, this study proposes a novel fire detection framework combining Multi-Feature Fusion and Adaptive Support Vector Machine (A-SVM). First, a high-dimensional feature vector is constructed by fusing HSI color space statistics, Local Binary Pattern (LBP) dynamic textures, and Wavelet Transform shape features. A baseline SVM classifier is then trained on source domain data. Second, to overcome the difficulty of acquiring labeled samples in target domains (e.g., strong daytime interference or low nighttime illumination), a small-sample domain adaptation mechanism is introduced. This mechanism fine-tunes the source model parameters using only a few labeled samples from the target domain via regularization constraints. Experimental results demonstrate that, compared with traditional color thresholding methods and unadapted baseline SVMs, the proposed method increases the F1-score by 19% and 30% in typical daytime and nighttime cross-domain scenarios, respectively. This study effectively achieves low-cost, high-precision, and robust cross-scenario fire detection, making it highly suitable for deployment on resource-constrained edge computing nodes within smart sensor networks. Full article

Analyzing fine-grained student actions across institutions can drive timely feedback, early warning, and personalized support, yet it is constrained by privacy regulations, heterogeneous curricula, and non-IID behavior logs. This paper introduces BAF–FedLLM, a behavior-aware federated modeling framework that adapts large language models to next-action and outcome prediction without centralizing student data. The key idea is to treat multichannel interaction streams as semantically typed action tokens linked by a learned ActionGraph, and to align their temporal structure with an LLM through behavior prompts that inject domain context (task, resource, pedagogy, and affordance cues). We propose three novel components: (i) BP–FIT, a behavior-prompted federated instruction tuning scheme that trains low-rank adapters locally and aggregates them with secure masking and Rényi–DP accounting to ensure client-level privacy; (ii) ProtoAlign, a cross-client prototype contrastive objective that shares only noisy class-conditional anchors via secure aggregation to mitigate drift under non-IID partitions; and (iii) CBR, a causal behavior regularizer that penalizes intervention-sensitive shortcuts by enforcing invariance of predicted risks across detected instructional regimes. We further derive convergence guarantees for federated instruction tuning with noisy, partial participation and provide end-to-end privacy bounds. On three public education datasets (EdNet, ASSISTments, and OULAD) with institution-level partitions, BAF–FedLLM improves next-action AUC by 4.2–7.1% over strong federated baselines while reducing expected calibration error by up to 28% and communication by $5\times$ through adapter sparsity, under a typical privacy budget of $\epsilon \approx 1.7$ at $\delta ={10}^{-5}$. These results indicate that behavior-aware prompting and prototype alignment make LLMs practical for privacy-preserving student action analysis at scale, offering a principled path to deployable, regulation-compliant analytics across diverse learning ecosystems. Full article

Community detection in graphs can be viewed as the estimation of a partition map that remains stable under admissible perturbations of graph topology and node attributes. While modern graph neural networks (GNNs) achieve strong empirical accuracy, they often exhibit severe assignment drift under minor perturbations, leading to illusory community structures. In this work, we propose DISPEL-GNN, a stability-aware graph learning framework that integrates spectral operator regularization, Bayesian uncertainty modeling, and risk-aware dynamic attention for perturbation-bounded community detection. The model explicitly constrains graph operators through uniform spectral norm bounds, high-frequency energy suppression, and commutator alignment while dynamically modulating message passing based on node-level spectral risk and epistemic uncertainty. We further formalize instability via assignment of drift functional and establish perturbation bounds linking drift to operator norms and spectral gaps, complemented by a PAC-Bayesian generalization guarantee. Extensive experiments on real-world benchmarks including Cora, Citeseer, Pubmed, Cora-Full, and DBLP demonstrate that DISPEL-GNN consistently reduces assignment drift by 18–35% under feature noise and edge perturbations while improving clustering quality with up to +3.0 NMI and +0.04 ARI compared to strong baselines such as GAT and Bayesian GNNs. The normalized mutual information (NMI), adjusted Rand index (ARI), and PAC-Bayesian (PAC) constraints serve as evaluative and theoretical instruments in this study. Additional studies on synthetic graphs with controlled spectral gaps confirm that the proposed method maintains stable community assignments in low-gap regimes where classical spectral and GNN-based methods degrade sharply. These results establish DISPEL-GNN as a mathematically grounded and practically effective framework for robust and interpretable community detection. A metric-wise dominance analysis shows that DISPEL-GNN achieves metric-wise dominance across most accuracy and robustness criteria, with minor tradeoffs in modularity on selected datasets. These results indicate that explicitly modeling stability and uncertainty provides a principled pathway toward reliable and interpretable community detection in noisy graph environments. Full article

This study addresses the challenge of suppressing radio frequency interference (RFI) in ultra-wideband (UWB) synthetic aperture radar (SAR) operating within complex electromagnetic environments, and proposes an innovative time–frequency signal extraction method. The proposed approach integrates a low-rank and sparse representation (LRSR) model in the time–frequency domain with a time total variation (TTV) constraint. The core contributions are twofold: (1) constructing a time–frequency LRSR model of frequency modulation continuous wave (FMCW) signal, and (2) incorporating spectral continuity as a prior via TTV regularization into a joint low-rank sparse optimization framework. This effectively reduces the aliasing of RFI components into the target components caused by improper hyperparameters, which is particularly pronounced under low signal-to-interference-plus-noise ratio (SINR) conditions. To enhance robustness, the incoherence of interference across frequency bands is exploited, and a sub-band coherence factor (CF) weighting technique is introduced to further suppress RFI residues in the image domain. Experimental results demonstrate that the proposed method significantly outperforms existing robust principal component analysis (RPCA)-based techniques, offering a more adaptive and robust solution for RFI mitigation in UWB SAR systems. Full article

This paper develops a framework for monitoring and forecasting episodes of systemic financial stress using a combination of market information, macro-financial indicators, and measures derived from time-varying correlation networks, embedded in a sequential machine-learning setting. The contribution is not tied to a single modelling innovation, but rather to the way these ingredients are brought together under an evaluation protocol designed to mimic real-time supervisory use, and to an interpretability layer that makes the resulting predictions easier to inspect. Monthly data covering the period from 2006 to 2025 are used to construct evolving correlation structures and summary indicators of market co-movement. These features are combined with standard predictors and fed into logistic regression, random forest, and gradient boosting models, all estimated in expanding windows and assessed strictly on future observations. Predictive accuracy remains limited, which is consistent with the difficulty of anticipating stress regimes several months ahead at monthly frequency, although gradient boosting attains the highest average AUC across evaluation folds and displays noticeable variation over time. Inspection of SHAP values points to instability in correlation networks, volatility conditions, and short-horizon return behaviour as recurring drivers of the predicted stress probabilities, suggesting that the models draw on information that goes beyond individual market series. Taken together, the results indicate that recurrent statistical regularities and changes in market structure can be exploited for monitoring purposes when models are trained and tested in a sequential fashion. The overall design is intended to be usable in practice and to support supervisory analysis, while remaining transparent enough to allow scrutiny of the signals driving the forecasts. Full article

Background/Objectives: Kenya’s diet-related non-communicable disease burden is rising alongside the consumption of ultra-processed foods. Kenya finalized a national nutrient profile model (KNPM) in 2025, drawing on the WHO African regional model (WHO NPM). The objective of this study was to examine labelling practices and the healthiness of packaged products available in Kenya, including domestically produced and imported items, to identify policy priorities to strengthen nutrient profiling, surveillance, and alignment with international standards. Methods: Packaged food and beverage data were obtained from Innova Market Insights. The proportion of products meeting minimum Codex nutrition labelling requirements was determined. The proportion of products that met the nutrient criteria set out under the KNPM (draft and final versions) and WHO NPM was examined overall and by category. Agreement between nutrient profile models was determined using Fleiss’ kappa. Results: Of 5587 products, 21% displayed minimum Codex nutrient requirements. Labelling was more complete among imported compared to domestic products (40% vs. 14%). Sales-weighted eligibility was low: 15% (WHO NPM and draft KNPM) and 17% (final KNPM). Agreement across models was 82% (k = 0.44) and was highest between the WHO NPM and the final KNPM (95%; k = 0.66). Beverage patterns reflected stricter thresholds in the WHO NPM and the final KNPM. Conclusions: Kenya’s packaged food supply is inadequately labelled, with a large proportion not meeting the nutritional requirements set out in the final KNPM or WHO NPM. Mandatory, Codex-aligned nutrition labelling is necessary to ensure full operationalization of the KNPM, with regular review to reflect evolving food environments. Full article

Background/Objectives: Nutrition labeling and food advertising are population-level strategies that may influence food choices. Excess weight is a recognized public health concern and a risk factor for cardiometabolic diseases; however, evidence regarding the association between label use, food advertising, and excess weight remains inconsistent. The objective of this study was to examine the associations between nutrition label reading and understanding, exposure to food advertising, food-related behaviors, and excess weight among Brazilian adults. Methods: A cross-sectional study was conducted with 580 adults living in the southern region of Minas Gerais, Brazil. Data were collected using a structured questionnaire addressing sociodemographic characteristics, food purchasing behaviors, exposure to food advertising, and habits related to reading and understanding nutrition labels. Excess weight was assessed using body mass index (BMI), calculated from self-reported weight and height. Logistic regression models and principal component analysis (PCA) were performed, adopting a significance level of 5%. Results: Excess weight was observed in 59.0% of participants. Regular use of nutrition labels was reported by 38.6% of respondents; among these individuals, 70.4% reported discontinuing the purchase of a food product after reading its nutritional information. In adjusted analyses, age over 30 years (p < 0.001), female sex (p = 0.006), higher number of dependents (p = 0.007), and type of media used (p = 0.005) were significantly associated with excess weight. The habit of reading nutrition labels was not independently associated with excess weight; however, better label understanding was associated with changes in food purchasing decisions. Considering the nutritional quality of foods as an important factor in food choices was associated with lower odds of having excess weight, although this association did not reach conventional levels of statistical significance (OR = 0.403; 95% CI: 0.15–1.00; p = 0.056). Conclusions: Excess weight among Brazilian adults was more strongly associated with sociodemographic and behavioral factors than with the habit of reading nutrition labels. Although nutrition labeling was not directly associated with excess weight, label understanding and perceived nutritional quality influenced food purchasing behaviors. These findings highlight the role of nutrition labeling and food advertising in shaping food choices and underscore the need for longitudinal studies to clarify their relationship with excess weight and related health outcomes. Full article

Ensuring food security and agricultural biosecurity increasingly depends on the rapid and accurate identification of harmful organisms that threaten crop production. Traditional identification methods rely heavily on expert knowledge, are time-consuming, and often fail in complex multi-species scenarios. To address these limitations, this study establishes a comprehensive image dataset that includes three major categories of agricultural harmful organisms—pests, weeds, and crop diseases—and proposes an enhanced convolutional neural network, DenseNet-CSL (DenseNet with Coordinate Attention, Deep Supervision, and Label Smoothing), developed based on DenseNet121 for efficient multi-class recognition. The dataset comprises 62 pest species, 28 weed species, and 30 major crop diseases, totaling 23,995 images collected under diverse growth stages, ecological conditions, and imaging environments. DenseNet-CSL incorporates three targeted improvements: a Coordinate Attention mechanism to strengthen spatial and channel feature representation, Deep Supervision to accelerate convergence and enhance generalization, and Label Smoothing Loss to regularize the output distribution and reduce overconfidence, which is beneficial under imbalanced and noisy data. Experimental results demonstrate that DenseNet-CSL achieves a precision of 81.3%, a recall of 80.1%, and an F1-score of 80% on the constructed dataset—outperforming DenseNet121, ResNet101, EfficientNetV2, and MobileNetV3—while shortening inference time by 1.36 s and adding only 1.772 MB of additional model parameters. These findings highlight the effectiveness of DenseNet-CSL for multi-class recognition of agricultural pests, weeds, and diseases, and underscore the importance of multi-source, multi-scene datasets for improving model robustness and generalization. The proposed framework provides a viable technical pathway for intelligent diagnosis and monitoring of agricultural harmful organisms, supporting port quarantine and agricultural biosecurity applications. Full article

Structural hyperparameter selection (HPS) in neural network (NN) regression faces two critical, computationally expensive barriers: the mandatory splitting of datasets for validation, which significantly impairs sample efficiency, and the inability of conventional metrics (like Data MSE) to decouple true modeling error from detrimental output noise, leading to suboptimal architectural complexity and overfitting. To resolve these systemic limitations, we propose the Reconstruction Error Minimization for Hyperparameter Selection (REM-HPS) framework, a novel, non-Bayesian approach grounded in statistical learning theory. REM-HPS fundamentally shifts the optimization objective by minimizing the Reconstruction Mean Squared Error (MSE), which precisely isolates and measures the model’s intrinsic ability to recover the underlying noise-free function. Since this target error is typically inaccessible, the framework employs the observable Data MSE (validation error) to construct a reliable, probabilistic estimate, yielding a deterministic and noise-aware selection criterion. REM-HPS utilizes a deterministic structural hyperparameter selection criterion that removes randomness due to validation data splitting, while remaining compatible with standard stochastic training procedures. This strategy allows for the use of the entire dataset for training, eliminating the need for explicit data splitting or the introduction of tuning-intensive regularization hyperparameters. Rigorous empirical validation demonstrates that REM-HPS consistently selects significantly more compact architectures (minimal complexity) while achieving superior generalizability and estimation accuracy, particularly across varied Signal-to-Noise Ratios and data regimes. By providing an efficient and optimal selection metric, REM-HPS offers a transformative, resource-efficient alternative to structural HPS in modern data-driven systems. Full article