Search Results (215)

The burgeoning scale of Pre-trained Large Models (PLMs) has intensified the demand for efficient inference without compromising performance, while existing large model collaborative frameworks have shown promise, they often suffer from functional redundancy among experts and limited robustness in complex cross-domain scenarios. In this paper, we propose Tri-gate Routing for Inference via Decoupled Efficient Network Technologies (TRIDENT), a highly efficient and robust heterogeneous collaborative inference framework. TRIDENT leverages the complementary inductive biases of MLP (for statistical patterns) and KAN (for symbolic logic) to maximize reasoning potential with minimal parametric overhead. To address feature homogenization in traditional distillation, we introduce Orthogonal Feature Decoupling Distillation, utilizing an orthogonality loss ${\mathcal{L}}_{orth}$ for functional decoupling and a reconstruction loss ${\mathcal{L}}_{recon}$ to anchor decoupled features to the PLM knowledge manifold. During inference, a Dual-Threshold Arbiter effectively detects expert hallucinations by integrating individual confidence ${\tau }_{con}$ and heterogeneous consistency ${\tau }_{agree}$. Extensive experiments on CIFAR-100-LT, XNLI, and GSM8K demonstrate that TRIDENT significantly reduces the Invocation Rate (IR) of PLMs while maintaining high accuracy. Our findings reveal a distinct Pareto optimal balance and validate the spontaneous division of labor between heterogeneous experts. By transcending the limitations of single-architecture systems, TRIDENT provides a robust and interpretable pathway for efficient collaborative intelligence. Full article

Low earth orbit (LEO) satellite orbit prediction is one of the key measures to compensate for position errors and ensure position accuracy, which plays an important role in the aerospace communication network for undertaking functions such as routing relay, real-time communication, and signal forwarding. However, existing learning-based satellite orbit prediction models that are recognized as the best measurement inevitably face the problem of distribution bias. Orbit predictions can lead to a decrease in model performance due to different types of satellites (LEO and SSO) and different time scales. In this article, a new method is explored to overcome these shortcomings. Unlike previous methods that explore the temporal correlation of orbit data, this novel orbit prediction method converts satellite orbit data into the frequency domain via Fourier transformation, using a third-order Fourier-derivative convolution framework. Specifically, the proposed Fourier dilation convolution (FDC) model demonstrates better generalization ability across different types of satellites and different time scales by combining frequency domain analysis and dilated convolution. Two real datasets are applied for experimental validation, and the results show the effectiveness of our proposed FDC model. Meanwhile, the proposed FDC model shows a decrease in mean absolute error (MAE) values compared to the temporal convolutional network based seasonal and trend decomposition using a Loess (STL-TCN) model. Quantitative comparisons demonstrate that compared to the STL-TCN model, the FDC model reduces the mean absolute error (MAE) by approximately 10% to 85% across different orbital dimensions. Finally, we conducted further analysis of the interpretability of the model. Full article

The integration of algorithmic systems into critical digital infrastructure is no longer peripheral to governance, it is governance. As AI-mediated decisions influence credit access, clinical diagnoses, criminal risk scores, and infrastructure routing, the question of who controls these algorithms and whether that control is meaningful has become a central concern for states and institutions at every level of development. Existing frameworks, including the NIST AI Risk Management Framework, ISO/IEC 42001, and the EU AI Act, have made real progress toward structured AI governance. However, none treats digital sovereignty as a first-order goal, nor do they provide integrated cross-layer guidance applicable across the diverse institutional landscape found worldwide. From this synthesis, we develop the Integrated AI Risk Governance Framework (IARGF): a four-layer structure covering policy and regulations, institutional oversight, technical controls, and operational execution, organized around five risk categories—technical, ethical, security, systemic, and sovereignty-related. A comparative analysis with major existing frameworks highlights the IARGF’s unique contributions, especially its explicit focus on sovereignty, adaptability across different institutional capacities, and recursive feedback mechanisms that connect all four governance layers. The framework is analyzed across three domains—healthcare AI, financial services, and critical infrastructure—to demonstrate its practical utility. Results confirm that governance effectiveness is a system property, not just a feature of individual layers; that digital sovereignty is both a governance goal and a distinct risk dimension with specific technical and institutional needs; and that context-aware, capacity-scaled governance is a design requirement, not a political compromise. The IARGF is presented as a conceptual governance model based on a systematic literature review rather than an empirically validated tool, and it remains to be tested in actual organizational settings. Its main contribution is the comprehensive theoretical integration of sovereignty, institutional capacity, and inter-layer governance dynamics, rather than proven performance advantages over existing models. Future research should aim to validate this framework through longitudinal case studies, expert panels, and retrospective failure analyses. Full article

Coherent optical communication serves as the backbone of long-haul, high-capacity optical networks, where polarization-insensitive 90° optical hybrids (OHs) are crucial for system simplification and robustness. This work presents a polarization-insensitive 90° OH based on asymmetric double-strip silicon nitride waveguides, designed for dual-polarization quadrature phase-shift keying (DP-QPSK) systems. The device consists of a cascaded polarization-insensitive structure incorporating one 1 × 2 and three 2 × 2 multimode interference (MMI) couplers, interconnected by four 90° bent waveguides. Optimized via 3D finite-difference time-domain (FDTD) simulations, the 1 × 2 MMI coupler exhibits insertion losses below 0.06 dB (TE) and 0.09 dB (TM), while each 2 × 2 MMI coupler shows insertion losses under 0.2/0.4 dB, amplitude imbalance below 0.05/0.18 dB, and phase error within ±0.5°/±1.5° for the TE/TM modes, respectively. Based on these components, the full device achieves polarization-insensitive operation across a 100 nm bandwidth (1500–1600 nm), with a phase error within ±1°, insertion loss below 0.3 dB (TE) and 0.5 dB (TM), and common-mode rejection ratio better than −40 dB (TE) and −30 dB (TM). Furthermore, the design demonstrates high fabrication tolerance, maintaining performance under manufacturing deviations of ±2 μm in MMI length and ±20 nm in waveguide spacing. This work provides a promising polarization-insensitive OH design and a viable route toward cost-effective mass production of next-generation high-speed coherent systems. Full article

A new series of eight novel etodolac-based 1,3,4-oxadiazoles was synthesized, characterized, and tested in silico in multidimensional routes, starting with etodolac, a well-known nonsteroidal anti-inflammatory medication (NSAID). In silico studies were performed prior to synthesis using the molecular docking technique in CCDC GOLD suite software (2025.3) to assess the interactions with two key targets involved in cancer pathogenesis: the crystal structure of the epidermal growth factor receptor EGFR tyrosine kinase domain (PDB ID: 4HJO) and the matrix metalloproteinase (MMP-9) complex (PDB ID: 5CUH). ADME studies were performed to assess the physicochemical properties of the synthesized molecules. Importantly, biotransformation prediction also indicated that the derivatives possess high metabolic stability, with hydroxylation of the thio-ether group as the primary predicted biotransformation route. All compounds were characterized using melting point, FT-IR, ¹H-NMR, and ¹³C-NMR spectroscopy. In vitro and/or in vivo experiments are needed to confirm this preliminary anticancer study. Full article

Conscious sedation is widely used in pediatric dentistry to manage dental anxiety, behavioral difficulties, and systemic diseases that compromise patient compliance with dental care. Despite its clinical importance, international recommendations vary considerably. Objective: To conduct a comparative analysis and evidence mapping of international clinical practice guidelines on conscious sedation in pediatric dentistry. Methods: A comparative guideline analysis and evidence mapping study was performed. Electronic searches were conducted in PubMed (MEDLINE), Scopus, EMBASE, Cochrane Database of Systematic Reviews, Web of Science, LILACS, SciELO, TRIP, and OpenGrey up to December 2023. Guidelines issued by recognized professional or governmental organization addressing conscious sedation in pediatric dentistry were included. Predefined domains were analyzed, including indications, contraindications, pharmacological agents, dosages, routes of administration, monitoring, discharge criteria, and professional training. Data were synthesized descriptively and graphically mapped to illustrate coverage patterns. Results: Twelve international guidelines were included. Complete convergence (100%) was observed in core safety domains, such as patient assessment, monitoring, and professional training. A high agreement was found for discharge criteria (91.67%) and contraindications (83.33%). However, substantial variability emerged in pharmacological protocols, with only 16.67% of guidelines providing comprehensive drug and dosage descriptions. Routes of administration and emergency equipment recommendations were inconsistently reported, appearing in 66.67% and 50% of guidelines, respectively. Conclusions: Although foundational safety principles are consistently addressed, significant heterogeneity persists in pharmacological and procedural recommendations. This variability may contribute to differences in practice and uncertainty among practitioners. Greater international harmonization of guidelines may improve consistency, enhance clinical decision-making, and strengthen patient safety in pediatric dental care. Clinical Relevance: Identifying areas of convergence and variability across international guidelines may support the development of more standardized sedation protocols and promote safer evidence-based clinical practice in pediatric dentistry. Full article

The present study demonstrates that the corrosion behavior of dental cobalt–chromium (Co–Cr) alloys is strongly influenced by the interaction between microstructure, manufacturing technique, and oral chemical environment. A comparative investigation was conducted on Co–Cr specimens fabricated using four technological routes: conventional casting, CAD/CAM machining, Selective Laser Melting (SLM), and Direct Metal Laser Sintering (DMLS). The study included microstructural characterization, evaluation of generalized corrosion behavior using the rotating electrode technique, assessment of localized crevice corrosion, and quantitative analysis of the release of twenty metallic cations. Extraction tests were performed for 168 h in two media simulating aggressive oral environments: 0.07 N HCl (acidic medium) and a fluoride-containing electrolyte (0.1% NaF + 0.1% KF). Electrochemical measurements were recorded in the current density range of 10⁻¹⁰ to 10⁻⁷ A/cm², while released cation concentrations were quantified at the µg/L level. All alloys exhibited very low corrosion current densities (icorr in the 10⁻⁸ to 10⁻⁹ A·cm⁻² range), confirming overall good corrosion resistance. Among all manufacturing routes, CAD/CAM specimens demonstrated the highest electrochemical performance, with a wide passivity domain extending up to approximately 740 mV/SCE. A statistical interaction analysis between extraction media and manufacturing techniques was performed using the non-parametric Mann–Whitney (MW) U test. Among the analyzed elements, only chromium showed a statistically significant difference between media (p < 0.05), with an approximately 25-fold-higher release in acidic conditions compared with the fluoride medium, confirming the predominant role of proton-induced destabilization of the protective Cr₂O₃ passive film. In contrast, fluoride-containing media induced selective release of elements such as Cu (3× higher), W (2.5× higher), and Mo (1.4× higher), associated with complexation phenomena. The manufacturing route significantly influences corrosion behavior. Although additive manufacturing technologies (SLM/DMLS) enable highly accurate and customized prosthetic designs, rapid solidification and microstructural heterogeneities may increase susceptibility to localized corrosion compared with more homogeneous CAD/CAM materials. Clinically, these findings suggest that future restorative strategies should incorporate corrosion-aware material selection within digital workflows. As digital dentistry evolves, predictive models integrating patient-specific oral conditions may assist clinicians in selecting the most appropriate material system for long-term performance. In conclusion, the long-term success of dental Co–Cr prosthetic devices depends not only on mechanical strength and precision of fit, but also on sustained electrochemical stability in the complex oral environment. Full article

Remote sensing images are highly susceptible to spatially non-uniform haze under complex atmospheric conditions, leading to contrast degradation and structural detail loss. Moreover, remote sensing scenes usually exhibit complex spatial structures, highly uneven haze distribution, and significant statistical variability, which further increases the difficulty of haze removal. To address this issue, we revisit the haze degradation mechanism of remote sensing imagery and propose a dynamic histogram-sorted Transformer dehazing method from the perspectives of statistical distribution modeling and region-adaptive restoration. Specifically, a Histogram-Sorted Adaptive Attention is designed to map spatial features into the statistical distribution domain through a dynamic histogram sorting mechanism, enabling explicit discrimination and precise modeling of regions with different haze densities. Meanwhile, a Perception-Adaptive Feed-Forward Network is constructed, which incorporates a stable routing-based mixture-of-experts mechanism to adaptively select restoration strategies according to local texture characteristics and global haze density, thereby significantly enhancing the adaptability of the model in complex remote sensing scenarios. Extensive experimental results demonstrate that the proposed method achieves superior performance over existing approaches across multiple remote sensing benchmark datasets, effectively improving both visual quality and robustness of remote sensing imagery. Full article

The rapid evolution of the Open Radio Access Network (O-RAN) architecture and the exponential growth in specification complexity create significant barriers for researchers translating 5G standards into practical implementations. Existing evaluation frameworks for large language models, such as ORAN-Bench-13K, focus predominantly on the theoretical comprehension of regulatory documents while neglecting the critical aspect of software execution. This disparity results in a profound semantic gap, defined here as the structural and conceptual misalignment between abstract normative requirements and their concrete realization in the source code of open platforms like srsRAN. To bridge this divide and enable advanced cognitive reasoning, this paper presents a Hybrid Retrieval-Augmented Generation (RAG) framework designed to unify two heterogeneous knowledge domains: the O-RAN/3GPP specification corpus and the srsRAN C++ codebase. The proposed architecture leverages a hierarchical Parent–Child Chunking strategy to preserve the structural integrity of complex code and normative protocols. Additionally, it introduces a probabilistic Semantic Query Routing mechanism that dynamically selects the relevant context domain based on query intent. This routing actively mitigates semantic interference—a phenomenon where merging conflicting cross-domain terminology introduces informational noise, which our baseline tests showed degrades response accuracy by 4.7%. Empirical evaluation demonstrates that the hybrid approach successfully overcomes this, achieving an overall accuracy of 76.70% and outperforming the standard RAG baseline of 72.00%. Furthermore, system performance analysis reveals that effective context filtering reduces the average response generation latency to 3.47 s, compared to 3.73 s for traditional RAG methods, rendering the framework highly suitable for real-time telecommunications engineering tasks. Full article

Chiral bound states in the continuum (BIC) metasurfaces have emerged as a promising platform for enhancing light–matter interactions, which have potential applications in advanced photonic and quantum information devices. However, simultaneously achieving near-perfect circular dichroism and highly efficient nonlinear conversion with highly symmetric structures in metasurfaces remains an open challenge. In this work, we design a $C_4$-symmetric chiral metasurface composed of eight elliptical silicon nanorods on a ${\mathrm{SiO}}_2$ substrate, where monocrystalline silicon is used as the nonlinear optical material. By combining simulations and nonlinear time-domain coupled-mode theory (TCMT), we discovered that both the optimal chirality and the nonlinear conversion efficiency can be attained simultaneously due to the critical coupling between the metasurface mode and the quasi-BIC mode. Meanwhile, a near-perfect circular dichroism (CD = 0.99) and a high nonlinear conversion efficiency of $7\times {10}^{-5}$ under a radiation intensity of $5{\mathrm{kW}/\mathrm{cm}}^2$ are numerically achieved due to the robustness of bound states in the continuum. This work offers a promising route toward high-performance chiral nonlinear photonic components, which is of great importance for the development of ultra-compact optical devices such as circular polarization detectors, chiral sensors, and nonlinear photonic chips for integrated optical and quantum information systems. Our research not only contributes to the fundamental understanding of chiral metasurfaces but also provides a practical approach for achieving high-efficiency nonlinear optical devices. Full article

Electroencephalography (EEG) has become an increasingly important tool in depression research due to its ability to capture objective neurophysiological abnormalities associated with depressive disorders, offering high temporal resolution, non-invasiveness, and cost-effectiveness.However, existing methods often fail to fully exploit the multi-domain information in EEG signals, resulting in limited model generalization capabilities. This paper proposes an improved bidirectional long short-term memory (BiLSTM) model that segments continuous EEG into non-overlapping 2-s epochs and learns end-to-end from multi-channel temporal sequences. After band-pass filtering and resampling, each epoch is represented as a channel–time matrix $X\in {\mathbb{R}}^{C\times T}$ (with C = 128) and processed by a BiLSTM encoder followed by a dynamic-routing encapsulated-vector classifier. On the MODMA dataset under subject-independent five-fold cross-validation, the proposed method outperforms a set of reproduced representative baselines (SVM, EEGNet, InceptionNet, Self-attention-CNN and CNN–LSTM) and achieves 84.8% accuracy with an AUC of 0.899. We further discuss recent contemporary directions (e.g., attention/Transformer-based and emotion-aware expert models) and clarify the scope of our empirical comparisons. Furthermore, experiments comparing different frequency bands and band combinations indicate that joint multi-frequency input can enhance classification performance. This study provides an effective multi-domain fusion approach for the automatic diagnosis of depression based on EEG. Full article

With the development of 6G communication technology, Space–Air–Ground Integrated Networks (SAGINs) have become critical infrastructure for global intelligent collaborative computing. However, federated learning deployment in SAGINs faces three severe challenges: the high dynamics of satellite orbital motion, node resource heterogeneity, and privacy vulnerabilities in data transmission. This paper proposes an adaptive compressed sensing differential privacy federated learning framework based on orbital spatiotemporal characteristics. First, we design orbital periodicity-driven time-varying sparse sensing matrices that dynamically adjust compression strategies according to satellite orbital positions, achieving intelligent communication efficiency optimization. Second, we propose an orbital predictability-based privacy budget temporal allocation mechanism and perform differential privacy noise injection in the compressed domain, establishing a compression–privacy joint optimization algorithm. Furthermore, we construct an energy–communication–privacy ternary collaborative mechanism that achieves multi-objective dynamic balance through model predictive control. Finally, we design reinforcement learning-based dynamic routing scheduling and hierarchical aggregation strategies to effectively handle the time-varying characteristics of network topology. Simulation experiments demonstrate that compared to existing methods, the proposed approach achieves 3–12% improvement in model accuracy and 30–50% enhancement in communication efficiency while maintaining differential privacy protection with dynamic privacy budget $\epsilon \in [0.1,10.0]$ and compression ratio $\rho \in [0.2,0.8]$. Unlike static compressed sensing approaches that ignore orbital periodicity, the proposed orbital-driven time-varying sensing matrices reduce reconstruction error by up to 19.4% compared to fixed-matrix baselines, validating the synergistic effectiveness of integrating orbital spatiotemporal characteristics with federated learning in 6G SAGIN deployments. The framework assumes reliable orbital propagation via SGP4/SDP4 models and does not account for Doppler frequency shifts or inter-satellite link handover delays; future extensions include scalability to mega-constellations and integration of quantum-resistant privacy mechanisms. Full article

The increasing availability of high-resolution gridded meteorological data poses significant challenges for efficient storage and rapid data access. This study proposes a route-aware adaptive variable-resolution storage (AVRS) strategy for gridded meteorological datasets. The spatial domain is partitioned into fixed-size blocks and storage resolution is dynamically assigned based on radar reflectivity characteristics and air-route traffic density, prioritizing aviation-relevant regions while reducing redundancy elsewhere. Composite radar reflectivity (CREF) data are used as a case study to evaluate storage efficiency, reconstruction accuracy, and query performance. Experimental results indicate that AVRS approach reduces storage volume while maintaining high reconstruction fidelity and preserving key convective structures. In addition, route-oriented point-based queries are significantly accelerated compared with conventional uniform-resolution storage. The proposed AVRS framework provides a scalable and aviation-oriented storage solution for large-scale gridded meteorological data, with potential benefits for atmospheric research and air traffic operations. Full article

Background: This paper investigates hybrid quantum–classical optimization approaches for addressing core supply chain management (SCM) problems. A unified hybrid framework is implemented and evaluated across five representative domains: vehicle routing, scheduling, facility location, inventory optimization, and demand forecasting. Methods: The framework integrates quantum algorithms—namely the Quantum Approximate Optimization Algorithm (QAOA), Quantum Annealing (QA), and the Variational Quantum Eigensolver (VQE)—with classical constraint-handling and local refinement procedures in an iterative workflow. Quantum solvers are employed for global solution exploration, while classical optimization ensures feasibility and convergence stability. Results: Experiments conducted on standardized synthetic benchmarks demonstrate that the proposed hybrid framework consistently outperforms classical-only and quantum-only baselines, achieving 12–18% reductions in operational costs and 20–35% faster convergence. In routing and fulfilment tasks, quantum-generated candidate solutions provide effective warm starts for classical refinement. Robustness analysis based on stochastic SCM simulations further indicates lower performance variance under uncertainty. Conclusions: These results demonstrate that hybrid quantum–classical optimization constitutes a practical and scalable strategy for near-term SCM decision-making under current Noisy Intermediate-Scale Quantum (NISQ) hardware constraints. Full article

Deploying Large Language Models (LLMs) on Maritime Autonomous Surface Ships (MASS) entails a critical trade-off between reasoning depth, inference latency, and hardware constraints. To fill the existing gap, we introduce MARTIAN (Maritime Agent for Real-time Tactical Inference And Navigation), a 14B-parameter decision support agent engineered for edge deployment on standard vessel hardware (e.g., the NVIDIA Jetson AGX Orin). Central to our approach is the Cognitive Core architecture, which utilizes a verified dataset of 21,800 Chain-of-Thought (CoT) instruction–response pairs to align general linguistic capabilities with maritime procedural logic. Empirical evaluations demonstrate that MARTIAN achieves an overall accuracy of 73.23% (SFT only) and 81.16% (SFT + RAG) on the Bilingual Maritime Multiple-Choice Questionnaire (BM-MCQ), a standardized assessment dataset constructed based on Officer of the Watch (OOW) competencies. Notably, the SFT-only configuration attains 78.53% on pure-logic-intensive COLREG tasks—surpassing the 72B-parameter Qwen-2.5 foundation model in this domain—while maintaining a real-time inference latency of 22.4 ms/token. Crucially, our ablation studies support a nuanced Interference Hypothesis: while RAG significantly enhances factual recall in knowledge-intensive domains (boosting total accuracy from 73.23% to 81.16%), it concurrently introduces semantic noise that degrades performance in pure logic reasoning tasks (e.g., COLREG maneuvering accuracy decreases from 78.53% to 77.36%). On the basis of this finding, we identify and empirically motivate a decoupled cognitive design principle that separates procedural reflexes (via SFT) from declarative knowledge (via RAG). While the full implementation of an adaptive routing mechanism is deferred to future work, the ablation results presented herein offer a validated, cost-effective reference architecture for deploying transparent and regulation-compliant AI on resource-constrained merchant vessels. Full article