Search Results (5,815)

Multimodal named entity recognition (MNER) aims to improve entity detection in social media texts by leveraging accompanying images, but its performance is often affected by weak text–image alignment, noisy or irrelevant visual content, and limited separation among entity representations. To address these issues, this study proposes CUA-MNER, a contrastive uncertainty–aware framework that combines hierarchical vision–text alignment, variational uncertainty–aware fusion, and token-level contrastive learning. The alignment module models correspondences at token, phrase, and sentence levels, allowing local visual regions and global image context to support textual entity recognition. The fusion module estimates epistemic and aleatoric uncertainty through variational inference and adaptively adjusts the contribution of each modality for different samples. The contrastive objective further encourages entity representations of the same type to be closer while separating different entity types. Experiments on the Twitter2015 and Twitter2017 benchmarks demonstrate that CUA-MNER achieves F1 scores of 76.97% and 89.66%, respectively, outperforming competitive baselines by 0.66 and 1.95 F1 points. Ablation and diagnostic analyses show that the three components provide complementary benefits. These results suggest that modeling modality reliability is useful for robust MNER, while the additional modules also introduce computational overhead and leave cross-domain generalization as an open issue. Full article

Efficient detection of aircraft surface defects (ASD) is a cornerstone of aviation safety. However, ASD detection is challenged by microscopic defect scales, extremely low contrast, and severe background interference. This paper proposes the Multi-Scale Discriminative Edge Enhancement Network (DEE-Net) based on an improved YOLO11. First, to mitigate feature dissipation of tiny defects, a lossless reassembly mechanism using space-to-depth convolution (SPD-Conv) is introduced, safeguarding sub-pixel topological information through space-to-depth conversion. Second, an adaptive selective edge-enhancement (ASE) module, integrating a dual-domain selection mechanism (DSM), is designed to suppress non-target redundant information on the fuselage skin. Finally, a Wise-CIoU loss function with a non-monotonic focusing mechanism is introduced to enhance localization stability under stringent IoU thresholds. Experimental results demonstrate that DEE-Net outperforms the baseline, improving mAP50 by 7.15% and mAP50-95 by 2.43%. To provide a more reliable evaluation, a 5-fold cross-validation experiment is further conducted on the original non-augmented images, and the results are reported as mean ± standard deviation. The cross-validation results provide a more conservative estimate and indicate that the proposed method achieves competitive performance across different data partitions. Full article

We propose a quantitative, spatially explicit framework for assessing local self-sufficiency and resilience within the Water–Energy–Food (WEF) Nexus. The methodology introduces normalized, per capita indicators that quantify the degree of dependence on local versus external resources, explicitly incorporating physical availability, renewability, energy requirements, infrastructure, and land-use constraints. In contrast to conventional composite indices, the proposed framework adopts a non-compensatory structure, whereby deficiencies in one sector cannot be offset by surpluses in another, reflecting the physical constraints of the nexus. Indicator values range from 0 (complete dependence on external resources) to 1 (full local self-sufficiency) and are formulated dynamically, enabling comparison across existing conditions and alternative infrastructural or policy scenarios. The framework is applied as a proof of concept to a small rural settlement in North Euboea, Greece. The results indicate substantial potential for food and renewable energy self-sufficiency under optimized infrastructure configurations, while also revealing critical vulnerabilities associated with groundwater-dependent water supply and seasonal energy imbalances. The analysis further demonstrates how spatial proximity, energy–water coupling, and land-use competition jointly constrain achievable self-sufficiency levels, highlighting trade-offs that are often overlooked in sectoral or purely volumetric assessments. By explicitly linking resource flows with spatial proximity and infrastructural choices, the proposed indicators provide a robust and transparent tool for resilience-oriented planning under conditions of climatic, environmental, and systemic uncertainty. Full article

This study investigated the impact of recreational use on trails in the Atlantic Forest (Ubatuba Municipality, São Paulo State, Brazil) using physical, chemical and geochemical indicators. Five trails with different morphological characteristics were selected, and paired samples were collected from the trail surface (TR) and trail-side slope (TA). The statistical approach combined local analyses for each trail with global clustering (n = 19) using Student’s t-test, along with multivariate modeling through Principal Component Analysis (PCA) and Pearson correlation. The analysis included physical attributes (bulk density, particle size and porosity), chemical attributes (pH, organic matter and macronutrients) and geochemical compositions (major oxides and trace elements determined by XRF). The overall results reveal systematic compaction in the trail surface (TR), with bulk density increasing from 1.32 g/cm³ (TA) to 1.37 g/cm³ (TR) (p = 0.038), and total porosity decreasing from 47.26% to 45.34% (p = 0.016). In contrast, the geochemical oxide composition (SiO₂, Al₂O₃, Fe₂O₃) remained stable (p > 0.05), indicating the resilience of the mineral matrix. However, significant local dynamics (p < 0.05) in K₂O and MgO were observed in more preserved trails, associated with surface compaction and fragmentation of the litter layer, and phosphorus showed strong dependence on organic matter (r = 0.85). Multivariate analysis indicates that degradation is predominantly physical and micromorphological at the local scale, with bulk density and porosity being the most sensitive indicators for environmental monitoring. Full article

In this paper, L1 phonemic systems are discussed in the context of their impact on the pronunciation and perception of English diphthongs in PE, drawing on World Englishes and phonetic analysis. The study focuses on speakers whose native languages are Punjabi, Seraiki, Pashto, and Urdu, and examines how changes in local vowel inventories and glide processes influence diphthong production. The controlled production and perception tasks were done on eight English diphthongs by 40 adult speakers (10 speakers per L1 group). The formant trajectories (F1, F2), duration, and intensity were recorded by acoustic analyses, which are used to measure the variation that occurs as the articulatory glide occurs between vowel targets. Perception was measured using diphthong identification tasks to assess listeners’ sensitivity to dynamic spectral movement. The results indicate systematic L1-conditioned restructuring. Deviations were the most pronounced in diphthongs with significant vowel gliding, especially centering diphthongs, characterized by a decrease in spectral movement, a constriction in vowel space, and a general tendency toward monophthongization. Closing diphthongs were generally more stable in production; however, they still exhibited systematic L1-conditioned variation, particularly in glide magnitude, spectral direction, and temporal realization. These patterns of production were highly consistent with the results of perceptual production: the diphthongs with lesser acoustic movement were also found to be less accurately recognized, and diphthongs in their L1s and speakers of phonemically richer vowel systems had partial glide contrasts. The findings demonstrate that the variation in diphthongs in PE is systematic, reflecting predictable relationships between the L1 phonemic system, perceptual assimilation, and sociolinguistic experience. The findings highlight the pedagogical value of L1-sensitive pronunciation instruction and contribute to the phonetic description of Pakistani English as a systematic contact variety. Full article

Partial occlusion can substantially impair the accuracy and stability of face recognition systems. Although existing methods perform well on unobstructed face images, their performance often degrades under partial occlusion, because occluded regions may obscure discriminative identity cues and introduce noise into feature representations. To address this issue, this paper proposes an occlusion-aware face recognition framework that integrates lightweight feature extraction, local-region reliability modeling, adaptive feature fusion, and joint loss optimization. Specifically, the face is divided into three sub-regions according to common occlusion patterns, and an MLP-based module is used to estimate the reliability of each region. The estimated reliability weights are then used to adaptively fuse local features, thereby emphasizing visible and discriminative regions. In addition, a joint loss combining ArcFace and InfoNCE is constructed to enhance inter-class separability and intra-class feature consistency. Experimental results under masks, hats, sunglasses, and random occlusion conditions show that the proposed method achieves a recognition accuracy of 92.3%. Compared with ArcFace, CurricularFace, and AdaFace, the proposed method improves accuracy by 9.9%, 6.5%, and 4.1%, respectively. In addition, the FAR is reduced by 5.8%, 4.9%, and 3.7%, respectively, while the FRR is reduced by 2.2%, 6.5%, and 1.3%, respectively. These results demonstrate that the proposed framework effectively enhances the robustness of face recognition under partial occlusion. Full article

This paper presents a comparative evaluation of quarter-car-model-based modular synthesis (QCMS) and full-car-based centralized synthesis (FCCS) for active suspension control in full-car systems. FCCS explicitly accounts for the coupled vertical, pitch, and roll dynamics by incorporating the geometric configuration of the sprung mass; however, this centralized formulation increases model complexity and controller–synthesis effort. In contrast, QCMS reduces the synthesis complexity by designing local suspension controllers using a quarter-car model and applying them modularly to the four suspension corners of a full-car system. Within both synthesis frameworks, linear quadratic (LQ) static output feedback (SOF) controllers and recursive-least-squares/extended-Kalman-filter (RLS/EKF)-based controllers are developed under comparable but structurally different control objectives. In particular, the proposed FCCS framework uses the geometric symmetry of the sprung mass not merely as a modeling assumption but as an explicit force-allocation structure that transforms the desired vertical force, roll moment, and pitch moment into four suspension actuator forces. Thus, four controllers are considered: LQSOF-QCMS and RLS/EKF-QCMS as modular quarter-car-based controllers, and LQSOF-FCCS and RLS/EKF-FCCS as centralized full-car-based controllers. In addition, the computational complexity of the LQSOF- and RLS/EKF-based controllers is compared in terms of their implementation burden. The main contribution of this study is not merely to show that the full-car-based FCCS improves the suppression of coupled body motions, but to clarify, under identical control and simulation conditions, the quantitative trade-off between the modular simplicity of QCMS and the symmetry-based centralized performance of FCCS. These controllers are evaluated through CarSim-based simulations under selected representative road-profile conditions in terms of ride comfort, motion-sickness mitigation, sensor requirements, and implementation complexity. The simulation results show that QCMS offers a low-complexity and modular implementation with acceptable ride-comfort performance, whereas FCCS justifies its increased synthesis and implementation burden when the suppression of coupled vertical, pitch, and roll motions and motion-sickness-related responses is required. Full article

Introduction: Understanding population structure is essential for effective fishery management, and otolith shape analysis provides a robust framework for detecting spatial variation in marine fish populations. The bluefish (Pomatomus saltatrix), a commercially important and widely distributed species along the southeastern and southern Brazilian coast, may exhibit subtle population structuring that is still not fully resolved. Methodology: Otolith contour variation was analyzed using two complementary approaches: Elliptical Fourier Descriptors (EFD) and Wavelet Transformed Descriptors (WTD). A total of 75 individuals (25/site) were sampled from Rio de Janeiro, São Paulo, and Santa Catarina, with total lengths ranging from 33.2 to 45.5 cm. Multivariate analyses included Permutational Multivariate Analysis of Variance (PERMANOVA), pairwise Hotelling’s t-tests (HT²), Flexible Discriminant Analysis (FDA), and jackknife reclassification matrices (JKC). Results: For the EFD approach, PERMANOVA showed no significant differences among localities, while FDA revealed partial overlap among groups and a JKC overall reclassification accuracy of 46%. In contrast, the WTD approach detected significant spatial differences, with PERMANOVA indicating overall variation among localities and HT² identifying significant differences between Santa Catarina and the other regions. FDA improved visual separation of Rio de Janeiro samples, although the JCK accuracy decreased to 35%. Conclusion: The combined results suggest the presence of weak to moderate spatial structuring in P. saltatrix along the studied coastline. However, inconsistencies among analytical approaches and relatively low reclassification success rate to the original site indicate that the observed differentiation is insufficient to conclusively define distinct population units, remaining compatible with either a single stock exhibiting spatial heterogeneity or weakly differentiated subpopulations. Full article

Introduction: Long-term ecological changes in estuarine communities are primarily driven by anthropogenic and environmental pressures. While abundance-based indicators are commonly used to assess these shifts, they often mask underlying ecological aspects related to age and/or size dynamics that may not necessarily be reflected in the abundance-based approach. Objective: This work tested a size-based indicator approach to examine the long-term changes in the size structure of the Mondego estuarine fish community (Portugal), using a 22-year dataset (2003 to 2025). Methodology: To capture the whole size structure, eight size-based indicators were applied, including mean length (MeanL), length at the 10th percentile (L10), median length (MedianL), length at the 90th percentile (L90), mean length of the 90th percentile (Lmax), size spectrum, the Large Fish index, and the Shannon index of length classes, at community and species levels and subsequently considered these in relation with with local and large-scale environmental factors. Results: Linear models identified a sharp, consistent decline in the overall size of the community, significantly correlated with the North Atlantic Oscillation index (NAO) and increasing estuarine water temperatures. A dynamic factor analysis (DFA) further identified one common trend across species for all indicators, corroborating the decrease in the overall size of the community while also acknowledging contrasting responses from different species, suggesting a heterogenous response across the fish community. Conclusions: These results highlight the importance of size-based indicators when assessing long-term ecological changes in marine ecosystems, allowing us to better understand how size structures shift, their relationship with a changing environment, and the long-term ecological outcomes in terms of community stability, resilience, recruitment, and ecosystem functioning. Full article

Progressive collapse has become a critical concern in resilient structural design due to accidental impacts, abnormal loading scenarios, and sudden localized damage events that may lead to the sudden loss of structural elements under extreme or unforeseen actions. In this context, UFC 4-023-03 provides design approaches for improving collapse resistance, including the Alternate Path Method, Enhanced Local Resistance Method, and Tie Forces Method. This study focuses on the Tie Forces Method, which is based on mechanical interconnection but remains relatively underexamined in the literature. A six-story reinforced concrete office building was evaluated to determine the required tie reinforcement area for progressive collapse resistance according to UFC 4-023-03. Ten national building codes were considered, with office live loads ranging from approximately 2.0 to 4.8 kN/m². In this study, the selected national codes are not compared in terms of their complete progressive collapse provisions. Instead, UFC 4-023-03 is adopted as the main Tie Forces Method calculation framework, while national-code-based live load values and reinforcement properties are used as input parameters. Peripheral, longitudinal, transverse, and vertical ties were comparatively assessed. The largest percentage reduction was observed for the peripheral transverse tie reinforcement at the first floor, where the Eurocode-based input set produced a required tie reinforcement area approximately 21.7% lower than that obtained from the Russian input set. In contrast, Canadian provisions govern the highest demand at the ground floor, while South Korean provisions produce the highest demand at upper floors. Overall, the findings highlight the influence of national live load provisions and reinforcement properties on tie force requirements. Full article

Background/Objectives: Spontaneous intracranial hypotension (SIH) is caused by spinal cerebrospinal fluid (CSF) leakage and is typically diagnosed by clinical presentation and characteristic MRI signs; however, objective tools for monitoring physiological changes and treatment response remain limited. Cine phase-contrast MRI (PC-MRI) enables noninvasive quantification of aqueductal CSF dynamics, yet reliable analysis is challenging since the cerebral aqueduct is extremely small and susceptible to low contrast, partial volume effects, and ROI-dependent measurement variability—particularly in SIH where CSF pulsatility is often reduced. Methods: We propose an end-to-end automated framework that integrates (1) a cascade localization–segmentation strategy, consisting of Tiny YOLOv4 detection followed by MultiResUNet segmentation on a YOLOv4-derived cropped ROI; (2) physiology-informed pulsatility-based segmentation (PUBS) to refine anatomical masks into functional flow ROIs; and (3) one-dimensional convolutional neural networks (1D-CNNs) to extract exploratory waveform morphology features from 32-phase cardiac-cycle velocity waveforms. The study includes 39 participants, yielding 59 cine PC-MRI examinations: 11 controls, 28 Pre-treatment SIH scans and 20 Post-treatment Recovery scans. Results: The cascade model significantly improves segmentation robustness compared with a full-image baseline, achieving higher Dice scores and markedly lower boundary errors across cohorts (e.g., Pre-treatment SIH HD95: 1.66 ± 0.74 px vs. 15.37 ± 44.98 px). PUBS refinement reduces quantification deviation from expert manual references in SIH (mean relative error: 7.4% to 5.6%) and improves diagnostic performance for multiple hemodynamic parameters (e.g., downward mean flow AUC: 0.747 to 0.792). For waveform morphology analysis, the end-to-end 1D-CNN classifier was evaluated using repeated-seed participant-level grouped LOOCV. The repeated-seed ensemble prediction showed modest out-of-sample discrimination between Normal controls and Pre-treatment SIH scans, with an AUC of 0.646, a bootstrap 95% confidence interval of 0.455–0.826, and a permutation-test p-value of 0.072. Separately, exploratory analysis of the final baseline-trained 1D-CNN latent space showed marked, apparent Normal-versus-SIH separability and an intermediate recovery distribution in PCA space, suggesting that aqueductal waveform morphology may encode SIH-related physiological information. Conclusions: These findings suggest that SIH-related information may be reflected not only in flow magnitude but also in aqueductal CSF waveform morphology. However, the modest and statistically non-significant out-of-sample performance of the end-to-end 1D-CNN classifier indicates that morphology-based AI features should currently be regarded as exploratory biomarker candidates rather than validated stand-alone diagnostic tools. Larger independent cohorts are required to confirm their reproducibility, physiological meaning, and clinical utility. Full article

Non-thermal millimeter-wave (MMW) irradiation represents a promising non-invasive strategy for cancer therapy, yet its effects in physiologically relevant 3D systems remain poorly defined. Here, we evaluated the biological impact of MMW exposure in 3D non-small-cell lung cancer (NSCLC) spheroids (NCI-H1299, A549) and normal WI-38 fibroblasts under active cooling to suppress bulk heating. We demonstrate that cellular responses are governed primarily by power density (PD), irradiation geometry, and genotype-dependent susceptibility. High-PD pyramidal horn (PH) irradiation (~4.9 mW/cm²) induced rapid apoptosis, metabolic collapse, and near-complete loss of clonogenic survival, whereas lower-PD waveguide (WG) irradiation (~0.6 mW/cm²) produced depth-limited, cumulative cytotoxicity. Surviving cancer cells exhibited robust senescence-associated growth arrest, particularly in p53-deficient NCI-H1299 cells, indicating a dual apoptotic–senescent anti-proliferative response. In contrast, WI-38 fibroblasts showed minimal apoptosis and only transient stress-associated senescence, confirming selective tumor vulnerability. Mechanistic modeling suggests that MMW energy couples to glycan-rich membrane domains, generating localized electromagnetic hotspots that trigger calcium influx, mitochondrial dysfunction, and depth-dependent apoptosis. These findings establish a mechanistic basis for selective, non-thermal MMW-induced cytotoxicity in 3D NSCLC models and support further preclinical development of MMW-based therapeutic strategies. Full article

Money laundering remains a critical challenge for financial systems because of the complex, hidden, and interlinked nature of illicit financial transaction networks. Understanding how these networks respond to targeted disruption is essential for exposing structural vulnerabilities and refining existing anti-money laundering (AML) prevention and intervention strategies. This study involves a social network analysis (SNA)-based resilience framework to evaluate the robustness of financial transaction networks through targeted node removal. In this approach, a network is represented as a directed graph, where nodes correspond to accounts and edges represent transactions. Centrality measures (i.e., degree, closeness, betweenness and pagerank), which capture local influence, global reach, and control over information flow, are applied to identify the most influential nodes. Network resilience is assessed by analyzing the variation in the size of the Largest Connected Component (LCC) under progressive node removal. An adaptive LCC-based resilience strategy is used, starting with large batches of nodes and gradually moving to smaller ones until the LCC drops below 50% of its original size, allowing for a more detailed analysis near the fragmentation threshold. The findings reveal that Betweenness centrality is the most effective metric in disrupting network connectivity under targeted attack scenarios, both outflow- and inflow-based analyses. Specifically, targeting only the top 2% of nodes by Betweenness centrality collapses the network’s core, reducing the Largest Connected Component (LCC) to 60% of its original size. In contrast, random attack strategy exhibit limited impact on overall network resilience compared to targeted approaches. Our findings provide actionable AML insights, showing that resilience-driven targeting of structurally critical accounts can effectively fragment money laundering networks and support more focused interdiction strategies. Full article

Six different twin boundary interface models were constructed by molecular dynamics simulations to investigate the effect of biaxial load ratio on the fracture behavior of titanium twin boundaries. Analysis of microstructural evolution indicates that twin boundaries exhibit a dual role during crack propagation. On one hand, they serve as preferential sites for void nucleation, promoting crack propagation along the twin boundary; on the other hand, they provide favorable sites for dislocation nucleation, inducing local plastic deformation at the crack tip, altering the crack path, and thereby hindering crack propagation. The crack propagation behavior in the ($\overline{1}0$11) and ($\overline{1}0$13) twin boundary models shows evident asymmetry: the crack on the left side mainly propagates through the void nucleation mechanism and exhibits a faster growth rate, while the right-side twin boundary inhibits crack propagation by favoring dislocation nucleation. In contrast, the crack propagation behavior in the ($\overline{1}0$12), ($\overline{2}1$11), ($\overline{2}1$12) and ($\overline{2}1$14) twin boundary models is largely symmetric on both sides, showing no significant difference in propagation rate. Stress field analysis further reveals that the differences in crack propagation behavior among the various twin boundary models mainly originate from the disparity in dislocation activity on both sides of the crack, resulting in different levels of stress concentration at the crack tip. When void nucleation occurs at the twin boundary interface, the stress concentration between the main crack and the void intensifies, promoting their coalescence and further propagation. Meanwhile, with an increase in biaxial load ratio, the stress concentration at the crack tip becomes more pronounced, further accelerating crack propagation. Full article

Persistent high-risk human papillomavirus infection is a critical prerequisite for cervical intraepithelial neoplasia and cervical cancer, yet viral factors alone do not fully explain why most infections clear while a subset persists and progresses. Emerging longitudinal, multi-omics, and mechanistic evidence supports a plausible model in which the cervicovaginal microbiota is not a passive bystander but a functional determinant of mucosal immunity, epithelial barrier integrity, and local metabolic tone. Lactobacillus-dominant community states, particularly those enriched in Lactobacillus crispatus, are generally associated with lower pH, regulated inflammatory signaling, stronger barrier function, and a higher likelihood of HPV clearance. In contrast, anaerobe-enriched dysbiosis is linked to elevated pro-inflammatory cytokines, altered antigen presentation, immune checkpoint signatures consistent with T-cell dysfunction, and metabolic shifts involving lactate depletion and accumulation of short-chain fatty acids and other metabolites that can influence epithelial and immune-cell programs. Importantly, the interaction is bidirectional: hrHPV can remodel the microenvironment by suppressing host defense peptides and perturbing mucosal barriers, thereby reducing Lactobacillus fitness and reinforcing dysbiosis in a feed-forward loop that favors persistence and oncogenic progression. This review integrates functional ecology, longitudinal clinical evidence, immunological and metabolic mechanisms, and translational implications, highlighting opportunities for microbiome-informed risk stratification and adjunctive interventions, as well as key gaps requiring standardized longitudinal multi-omics and rigorously designed clinical trials. Full article