Search Results (3,277)

Risk-based prioritization frameworks such as the Intelligent Pedestrian Model (IPM) rank pedestrians by an instantaneous, reference-normalized risk score. They indicate which pedestrian requires attention first. This study examines the temporal stability of such rankings. We computed an observation-only kinematic Exposure proxy frame by frame on three ETH/UCY benchmark scenes. In these scenes, the highest-priority identity changes rapidly: the median top-1 persistence is two frames (0.8 s). We introduce a switch classification that separates established switches from entry-driven and forced switches. Grace-period exclusion is evaluated as a sensitivity variant and shown to remove up to 82% of evaluable time in short-track scenes. Established flicker rates range from one switch per 2.7 s in dense scenes to one per 25.4 s in sparse scenes, with established switches concentrated at small score differences between competing pedestrians. The results show that instantaneous rankings alone may be insufficient for sustained attention allocation and motivate future work on temporal priority management. Full article

Online social networks generate large volumes of textual data that reflect users’ opinions, affective expressions, and broader patterns of engagement and social behavior. However, natural language processing approaches frequently examine sentiment, trust-related signals, and behavioral indicators independently, limiting their ability to represent the multidimensional nature of online interaction. This study conducts a systematic comparative evaluation of lexicon-enhanced fine-grained sentiment classification using linguistic, message-level statistical, and lexicon-derived affective information across a common experimental framework. The empirical analysis combines TF–IDF features, word-count information, and sentiment indicators derived from TextBlob, SentiStrength, and VADER, while the broader multi-level organization is used to relate the resulting affective evidence to online social-behavior analysis. Fifteen classical machine learning algorithms and seven deep learning architectures are evaluated on a real-world Twitter dataset containing 41,157 COVID-19-related tweets labeled across five sentiment-intensity classes. The experimental evaluation considers four feature configurations and seven performance metrics, complemented by Friedman and post hoc Wilcoxon signed-rank tests. The results show that TextBlob provides modest improvements, SentiStrength produces broader and more consistent gains, and VADER yields the strongest overall performance. AdaBoost combined with VADER achieves the best results, with 93.16% accuracy, 93.20% macro F1, 93.27% balanced accuracy, and an MCC of 0.913, while the Dense Neural Network is the strongest deep learning model. These results demonstrate that lexicon-derived affective features can substantially strengthen fine-grained sentiment classification, although their effectiveness depends strongly on the learning algorithm used to exploit them. The empirical contribution of this study is confined to fine-grained sentiment classification, while trust-related and attachment-related dimensions are retained as higher-order interpretive constructs rather than directly predicted or empirically validated outcomes. Full article

Pilgrim safety during Hajj is challenged by dense crowds, sustained physical exertion, and demanding environmental conditions that can increase tiredness and compromise well-being. Although wearable sensing enables objective monitoring, the reviewed literature provides limited evidence on how synthetic augmentation affects predictive performance and cross-participant generalization when wearable data are scarce. To address this gap, this study presents an integrated framework that combines correlation- and domain-informed feature screening with statistically constrained synthetic data generation and a complementary assessment of predictive performance and synthetic data fidelity. Using the Hajj 1445 (June 2024) Apple Watch dataset comprising 120 records from three participants, the study compares classical resampling, univariate normal generation, multivariate normal (MVN) sampling, and covariance-aware generation using Cholesky decomposition and Mahalanobis distance filtering under a consistent classifier evaluation protocol. The highest observed record-level test accuracy is 93.33%, achieved by a Decision Tree using MVN sampling with moderate clipping ($\pm 1.4$). Leave-one-participant-out evaluation of the same configuration yields a mean accuracy of 68.65% (SD = 19.72%), indicating lower cross-participant generalization. The regularized MVN–Cholesky–Mahalanobis configuration attains a composite synthetic-data quality score of 79.27%. These results show that statistically constrained synthetic augmentation can support tiredness prediction when real data are limited, while the participant-grouped results highlight the need for validation using larger and more diverse Hajj cohorts. Full article

The recent trend of Software-Defined Networking (SDN) has posed significant cybersecurity challenges as a result of its centralized control architecture, dynamic traffic behavior, and high programmability. Although these attributes improve network flexibility and management, they also increase vulnerability to Distributed Denial-of-Service (DDoS) attacks that can overwhelm network resources and disrupt services. Traditional signature- and rule-based detection methods may struggle with evolving traffic patterns and generate excessive false alarms. Machine learning offers a more promising solution that can learn the complex traffic patterns and separate malicious traffic from normal traffic. Most machine learning models, however, are black-box models that provide only superficial insight into the model predictions. Explainable Artificial Intelligence (XAI) addresses this limitation by identifying influential traffic features and providing interpretable evidence for detection decisions. This research develops an explainable machine learning-based framework for accurate, transparent, and reliable DDoS attack detection in an SDN environment. Several machine learning models are assessed, and XAI techniques are applied to explain the results of the predictions at global and instance levels. Gradient Boosting, Logistic Regression, AdaBoost, and Gaussian Naive Bayes were evaluated on 104,345 network-flow records using a 70:30 training–testing split. Gradient Boosting achieved the strongest performance, with 99.88% training accuracy, 99.87% testing accuracy, a testing F1-score of 99.84%, and a 0.20% miss rate. SHAP identified the most influential traffic features, while LIME linked individual predictions to feature-specific contributions. The proposed framework therefore combines reliable DDoS detection with transparent, analyst-oriented decision support for SDN security monitoring. Full article

In the field of cybersecurity, malicious website classification plays a crucial role in protecting industrial systems. For this reason, research has been undertaken to analyze cybersecurity threats, with the long-term objective of developing methods for the effective detection and classification of malicious websites. This article evaluates the use of domain features for classifying malicious websites with machine learning methods. The feature vector consisted of 44 infrastructural, lexical, structural, and reputation-related characteristics. The model comparison included Logistic Regression, Support Vector Machine, Random Forest, AdaBoost, and XGBoost. Experiments were conducted on 247,730 URLs from the malicious_phish dataset (2021), with features extracted as part of this study in April 2026. The most predictive features were related to domain registration history, DNS infrastructure, and reputation-based rankings, as confirmed by ANOVA F-test, SHAP values, XGBoost gain, and permutation importance. Validation of the best-performing model on 1000 active phishing domains from the PhishDestroy list dated 30 May 2026, achieved a recall of 70%, while the application of a three-tier risk scale allowed 84.9% of domains to be flagged as malicious or suspicious. Full article

Due to industrialization, urbanization, and the need for sustainable water resources, the treatment of organic water has become a major environmental concern. Semiconductor photocatalysts degrade organic contaminants through a clean, energy-efficient process called photocatalysis. In this study, batch reactor experiments were performed under natural solar irradiation to examine the influence of zinc oxide (ZnO) dosage, pH, and reaction time (inputs) on photocatalytic performance, with total organic carbon (TOC) as the output. Three machine learning algorithms, namely Artificial Rabbit Optimization–Support Vector Regression (ARO–SVR), Support Vector Regression (SVR), and AdaBoost, were proposed and compared. Model performance was assessed by the coefficient of determination (R²), root mean squared error (RMSE), mean absolute error (MAE), and mean squared error (MSE). The proposed ARO–SVR model outperformed the other models, with training and testing R² values of 0.9782 and 0.9731, RMSE values of 0.040 and 0.0453, MAE values of 0.0402 and 0.046, and MSE values of 0.0016 and 0.0020. By contrast, conventional SVR and AdaBoost showed lower prediction accuracy and larger testing errors, indicating a lower generalization ability. The results show that combining photocatalytic technology in seawater treatment with an optimized machine learning framework can reliably and efficiently predict seawater treatment performance and minimize the need for extensive lab experiments. Full article

Lane-Keeping Assist Systems (LKAS) are widely used in modern passenger vehicles to support lateral vehicle control and reduce the risk of unintended lane departure. Although LKAS performance is commonly evaluated in relation to perception, control, and sensor fusion, the observable vehicle response may also depend on the mechanical condition of the chassis. This study presents a qualitative, measurement-based evaluation of the influence of intentionally introduced front-wheel toe misalignment on the observable response of a production LKAS under controlled proving-ground conditions. Experimental tests were conducted on the highway module of the ZalaZONE proving ground using a Lexus RX 450h equipped with a factory-installed LKAS function. Three front-wheel toe configurations were investigated: factory-specified alignment, single-wheel toe misalignment, and severe toe misalignment affecting both front wheels. Measurements were performed at 70, 90, and 110 km/h on straight and curved road sections. Vehicle speed, steering angle, lateral acceleration, and GNSS-based position data were recorded using CAN- and GNSS/IMU-based data acquisition. The qualitative comparison of the measured signal profiles indicated that the misaligned configurations were associated with a shifted steering-angle operating range and less uniform steering and lateral-acceleration responses. The most pronounced visible differences occurred under the severe toe-misalignment condition, particularly at higher speeds and in the curved section. As the analysis did not include quantitative effect measures or statistical comparisons, these observations are interpreted as exploratory tendencies rather than statistically validated changes in LKAS performance. The findings suggest that front-wheel toe condition should be considered in the measurement-based assessment, maintenance, and calibration of ADAS-equipped vehicles. Full article

Nighttime driving safety remains a critical challenge in modern transportation systems: insufficient ambient lighting significantly degrades visual perception quality, adversely affecting both human drivers and advanced driver-assistance systems (ADAS) and directly threatening road users’ safety. Traditional image enhancement methods often suffer from color distortion and visual artifacts, whereas existing deep learning approaches typically require paired training data and incur substantial computational overhead. To address these limitations, this paper presents BLEN (bio-inspired low-light enhancement network), a zero-reference deep learning framework that integrates biological vision principles with efficient convolutional architectures. Specifically, BLEN leverages Retinex theory for illumination–reflectance decomposition, is inspired by and functionally approximates lateral inhibition mechanisms for edge enhancement, and incorporates a Large-Kernel Convolution with Attention (LKCA) module that reduces the parameter count of the LKCA encoder block by 76% (0.56 M vs. 2.34 M for a standard 13 × 13 convolution) relative to standard large-kernel operations. Extensive experiments on the SICE and LOL benchmarks demonstrate that BLEN achieves state-of-the-art performance among real-time, edge-deployable zero-reference methods on the SICE benchmark, yielding a peak signal-to-noise ratio (PSNR) of 23.67 ± 0.14 dB and a structural similarity index measure (SSIM) of 0.891 ± 0.004 on SICE while maintaining 2.10 M parameters (2.1 MB in INT8, 8.4 MB in FP32). Furthermore, the proposed method enables real-time inference at 31 frames per second (FPS) on embedded platforms, including the HiSilicon SS928 and Jetson Nano, demonstrating that the proposed method is an efficient and effective front-end for camera-based ADAS perception on automotive-grade edge hardware. Full article

Early dropout prediction requires models that integrate heterogeneous educational records without information leakage. We propose multi-view neural risk augmentation (MVNRA), a cross-fitted framework that separates early-warning predictors into academic, contextual, and digital-engagement views. View-specific encoders and a gated fusion module produce four supervised risk scores. We evaluated MVNRA on a public longitudinal dataset from a Spanish technological university containing 464,739 student–course records and 81 predictors available by December. Within each outer student-level GroupKFold split, the neural risk generator was trained through inner student-grouped cross-fitting. Its out-of-fold fused, academic, contextual, and digital risk scores were then appended to the original predictors before six conventional classifiers were fitted. MVNRA increased mean area under the receiver operating characteristic curve (AUC) for all six classifiers. The largest gains occurred for AdaBoost (0.7717 to 0.9240), linear discriminant analysis (0.8701 to 0.9176), Gaussian Naive Bayes (0.7313 to 0.7786), and Logistic Regression (0.9139 to 0.9352). MVNRA + Random Forest achieved the highest absolute discrimination (AUC $=0.9577$), with increases of 0.0191 in precision–recall AUC and 0.0112 in F1 score. SHAP analysis ranked the fused and academic neural risk scores above the cumulative academic-progress indicators. These findings show that cross-fitted, view-specific neural risk scores can transfer nonlinear structure to conventional tabular classifiers while preserving a SHAP-compatible decision layer and strict student-level separation. Full article

In this work, it has been demonstrated that the use of hydrogen bond acceptors (HBAs) within the context of deep eutectic solvents (DESs) can be beneficial for fructose dehydration in ethanol to obtain high yields of 5-hydroxymethylfurfural (HMF) and 5-ethoxymethylfurfural (EMF). In this research, several commercial catalysts were screened in the dehydration of fructose under reflux in ethanol. Sulphonated polymers exhibited excellent performance, while zeolites only achieved low yields of HMF due to low acidic strength, rendering them unable to perform the dehydration and etherification of fructose and its subsequent products in the assayed conditions. The addition of an HBA in the form of tetraethyl ammonium bromide (TEAB) was essential to facilitate the dehydration of fructose with zeolites (up to 19% HMF yield with TEAB). In addition, the use of TEAB improved the observed yield of HMF twofold during the reaction when the sulphonated polymer Purolite PD206 was employed (34% with TEAB and 15% HMF yield without TEAB). The investigation of different HBAs showed that tetraethylammonium bromide (TMAC) produced the most promising results, achieving a furanic compound yield of up to 75% after 3 h of reaction at 120 °C, enhancing fructose dehydration kinetics and favoring the stability of furanic rings in the reaction medium. Full article

Flood forecasting often depends on upstream hydrographs that are unavailable, incomplete, or unreliable. Reverse Flood Routing (RFR) addresses this gap by reconstructing upstream inflows from downstream observations, yet its operational use remains limited by numerical instability, observational uncertainty, and the ill-posed nature of the inverse problem. This review critically synthesizes RFR methodologies across a physics–fidelity continuum, ranging from storage-based and simplified hydraulic models to full hydrodynamic inversions, optimization-based techniques, Bayesian approaches, and emerging data-driven methods. The reviewed approaches are compared in terms of physical realism, numerical stability, computational demand, data requirements, uncertainty treatment, and field applicability. The synthesis indicates that storage-based methods remain attractive for data-limited and computationally constrained applications, whereas full hydrodynamic models are better suited to complex flow conditions involving backwater effects and detailed channel hydraulics. Optimization-based and Bayesian approaches can improve parameter estimation and uncertainty representation, while hybrid AI–physics methods offer promise for computational acceleration but still require stronger physical constraints and broader operational validation. Across all methodological families, error amplification, lateral inflow, transmission losses, and inconsistent benchmarking remain persistent limitations. An integrated framework is therefore proposed to connect observations, model selection, regularization, uncertainty quantification, hybrid computational methods, and operational decision support, providing a roadmap for more reliable and scalable RFR applications. Full article

Hyperuricemia (HUA) is a metabolic disorder caused by excessive uric acid (UA) production and/or insufficient UA excretion. Brassica rapa L. (turnip) is a traditional medicinal and edible plant in northwestern China. In this study, the urate-lowering effects of B. rapa L. ethanol extract (BrEE) and associated underlying mechanisms were investigated using a hypoxanthine/potassium oxonate-induced HUA mouse model and UA-stimulated HK-2 cells. BrEE was prepared and characterized; quercetin-3-glucuronide was the most abundant compound in BrEE, with a relative abundance of 22.04%. BrEE significantly lowered serum UA levels and improved renal function in mice. Key UA-producing enzymes, xanthine oxidase (XOD) and adenosine deaminase (ADA), were suppressed and renal UA transporters were modulated to enhance excretion: URAT1 and GLUT9 were downregulated and ABCG2 was upregulated. These results indicated that BrEE’s protective effects were associated with its regulatory effects on UA transport-related protein expression. BrEE also alleviated oxidative stress, and reduced inflammation and renal pathology. Meanwhile, molecular docking predicted favorable binding between the major BrEE constituents and XOD or URAT1, providing preliminary support for their potential interactions with these targets. Thus, BrEE exerted multi-target urate-lowering and kidney-protective effects, providing preliminary evidence for the potential of B. rapa L. as a plant-derived resource for further investigation in the development of functional food ingredients to regulate urate metabolism. Full article

Background: Over the last decade, numerous studies have investigated the administration of monoclonal anti-amyloid antibodies (AAAs) as a therapeutic approach in Alzheimer’s disease (AD). The purpose of this umbrella review is to summarize current knowledge concerning the efficacy and safety of FDA- and EMA-approved AAAs for AD, namely, lecanemab and donanemab. Methods: We conducted a literature search in the PubMed, Scopus, and Web of Science databases in English, focusing on systematic reviews and/or meta-analyses, assessed by AMSTAR 2, concerning clinical and imaging efficacy, i.e., cognitive improvement and reduction of beta-amyloid on PET scan, as well as safety, i.e., adverse events and amyloid-related imaging abnormalities (ARIA). The extracted review-level dataset was entered into SPSS Version 26 for descriptive statistical analysis. Results: This umbrella review included 11 systematic reviews and/or meta-analyses. Patients treated with lecanemab or donanemab showed statistically significant changes on cognitive scales such as the Clinical Dementia Rating-Sum of Boxes (CDR-SB) (mean effect estimate −0.485, SD = 0.152, −0.70 to −0.34) and the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), but most were below or around clinically meaningful thresholds. Lecanemab and donanemab showed a statistically significant decrease in amyloid PET outcomes, including centiloids or the standardized uptake value ratio. Both antibodies were linked to ARIA, including amyloid-related imaging abnormalities-edema (ARIA-E OR 8.32–12.26) and amyloid-related imaging abnormalities-hemorrhage (ARIA-H OR 2–5.77), especially in ApoEε4 carriers. Conclusions: Lecanemab and donanemab are biologically active drugs for the treatment of early AD, with cognitive benefits below or around the clinically meaningful threshold. They are, however, associated with increased ARIA risk. Their administration depends upon careful patient selection, shared decision-making, and clear communication regarding expectations. Full article

Electroencephalography (EEG)-based emotion recognition provides an important physiological basis for affective brain–computer interfaces. However, high-dimensional EEG features vary across trials, sessions, and subjects. Euclidean distance and conventional source-only LMNN cannot jointly learn a discriminative geometry and align cross-domain emotion distributions. To address this problem, we propose Adaptive Large Margin Nearest Neighbor (AdaLMNN), an unsupervised domain-adaptation framework for EEG emotion recognition. AdaLMNN learns a shared Mahalanobis projection through normalized large-margin discrimination, adaptive margins, and refreshed source neighborhoods. It further uses confidence-guided target pseudo-labels, class-balanced conditional alignment, and scale regularization with scheduled optimization to reduce source–target mismatch without using target ground-truth labels. We evaluated AdaLMNN on synthetic data and the SEED and SEED-IV datasets using SVM, KNN, and MLP under three protocols. AdaLMNN substantially improved EEG emotion-classification performance, particularly under within-session and session-level shifts. Although cross-subject gains remained dataset- and classifier-dependent, the results demonstrate the potential of AdaLMNN as an interpretable and extensible representation-learning module for EEG emotion recognition under heterogeneous recording conditions. Full article

Crataegus L. (Rosaceae) comprises several species rich in nutrients and bioactive compounds with nutraceutical relevance. The present study aimed to investigate the polyphenolic profile, antioxidant activity, and in vitro antihyperglycemic-related activity of fruits from 11 Crataegus taxa collected in Western Sicily, a region characterized by a high diversity of this genus. Total polyphenol quantification indicated that, among the tested extracts, Crataegus laciniata exhibited the highest content (25.54 mg GAE/g). Using high-resolution Orbitrap Q-Exactive PLUS ESI-MS instrument (Thermo Fisher Scientific, Waltham, MA, USA), a comprehensive phytochemical characterization was performed. Hyperoside quantification revealed that Crataegus aff. insengae, Crataegus zichichii, and Crataegus drepanensis are rich in this bioactive compound (1.47, 1.22, 1.22 mg/g DW). Multivariate analysis, including principal component analysis (PCA), highlighted hyperoside and vitexin as the main contributors to sample differentiation. Additionally, Crataegus laciniata, Crataegus drepanensis, and Crataegus aff. insengae showed the highest antioxidant activity in chemical assays (DPPH, ABTS, and FRAP) and in Caco-2 and H9c2 cell models through the modulation of intracellular reactive oxygen species (ROS). Regarding antihyperglycemic-related activity, Crataegus laciniata, Crataegus azarolus var. azarolus, and Crataegus drepanensis extracts showed the best inhibitory activity against α-glucosidase and α-amylase. Overall, the study highlights the value of Crataegus fruits as sources of bioactive compounds for food and nutraceutical research and supports their further valorization. Full article