Search Results (24,300)

Fully homomorphic encryption (FHE) offers a promising solution for privacy-preserving machine learning by enabling arbitrary computations on encrypted data. However, the efficient evaluation of non-linear functions—such as the ReLU activation function over large integers—remains a major obstacle in practical deployments, primarily due to high bootstrapping overhead and limited precision support in existing schemes. In this paper, we propose $\mathsf{LargeIntReLU}$, a novel framework that enables efficient homomorphic ReLU evaluation over large integers (7–11 bits) via full-domain bootstrapping. Central to our approach is a signed digit decomposition algorithm, $\mathsf{SignedDecomp}$, that partitions a large integer ciphertext into signed 6-bit segments using three new low-level primitives: $\mathsf{LeftShift}$, $\mathsf{HomMod}$, and $\mathsf{CipherClean}$. This decomposition preserves arithmetic consistency, avoids cross-segment carry propagation, and allows parallelized bootstrapping. By segmenting the large integer and processing each chunk independently with optimized small-integer bootstrapping, we achieve homomorphic ReLU with full-domain bootstrapping, which significantly reduces the total number of sequential bootstrapping operations required. The security of our scheme is guaranteed by TFHE. Experimental results demonstrate that the proposed method reduces the bootstrapping cost by an average of 28.58% compared to state-of-the-art approaches while maintaining 95.2% accuracy. With execution times ranging from 1.16 s to 1.62 s across 7–11 bit integers, our work bridges a critical gap toward a scalable and efficient homomorphic ReLU function, which is useful in privacy-preserving machine learning. Furthermore, an end-to-end encrypted inference test on a CNN model with the MNIST dataset confirms its practicality, achieving 88.85% accuracy and demonstrating a complete pipeline for privacy-preserving neural network evaluation. Full article

This paper presents an integrated mathematical model to improve the safety and operational efficiency of train traffic in centralised railway dispatching systems. The proposed approach combines the alternative graph model with a Mealy automaton to synchronously address route planning, delay minimisation, and strict compliance with safety requirements. Formal control theory based on finite-state automata is employed to describe routing logic and signal control through state transitions, while the alternative graph model represents scheduling constraints and resource conflicts. To enhance real-time adaptability, a tabu search algorithm is implemented for train schedule optimisation, enabling dynamic rescheduling under changing operational conditions. The mathematical formulation incorporates blocking time parameters, a system of discrete constraints, and automaton-based safety conditions governing train movements and route authorisation. The integrated model explicitly formalises the processes of block section occupation and release, ensuring consistency between control logic and scheduling decisions. Practical testing and computational experiments demonstrate that the proposed approach effectively reduces train delays, improves the reliability of dispatch control, and increases system resilience to dynamic disturbances. The results confirm that the developed model can be implemented within existing centralised dispatching infrastructures without requiring a complete system overhaul. Overall, the proposed framework expands the functional capabilities of centralised dispatch systems by enabling efficient schedule generation, minimising the propagation of delays, and ensuring reliable command exchange between central control posts and field-level railway infrastructure. Full article

Long-term operation of drip emitters under sediment-laden water conditions readily induces particle deposition and clogging, leading to discharge reduction and deterioration of irrigation uniformity. To clarify the temporal evolution and spatial distribution of clogging and to support structure-oriented anti-clogging improvement, three integrated drip tape emitters with different labyrinth-channel geometries were tested at sediment concentrations of 1, 2, and 3 g·L⁻¹ under a constant pressure of 100 kPa. The average relative discharge ratio (Dra) and Christiansen’s uniformity coefficient (CU) were continuously monitored, and cross-sectional observation and numerical simulation were combined to identify dominant deposition hotspot regions within the labyrinth channel. The results showed that increasing sediment concentration significantly accelerated clogging development and shortened operating lifetime. At 1 g·L⁻¹, the times required for the three emitter types to reach the clogging criterion of Dra < 75% were 120, 81, and 107 h, respectively, whereas at 3 g·L⁻¹ these values decreased to 39, 42, and 39 h. CU continuously declined with operating time and, in some treatments, responded earlier than Dra to system deterioration. Sediment deposition was mainly concentrated in the inlet section and bend regions, indicating that these locations were the dominant hotspots for clogging initiation and propagation. These findings demonstrate that clogging in drip emitters is jointly regulated by sediment load and labyrinth-channel geometry, and that hotspot-based structural optimization provides an effective basis for improving anti-clogging performance under sediment-laden water conditions. Full article

This study systematically investigated the bearing capacity and failure mechanisms of ultra-high-performance concrete (UHPC) pipe jacking structures using three-edge bearing tests and numerical simulations. Full-scale double-layer reinforced pipes had an inner diameter of 2.5 m and wall thicknesses of 180 mm (P1) and 200 mm (P2). The tests showed that the failure process can be divided into four stages: elastic deformation, crack propagation, reinforcement yielding, and ultimate failure. Increasing the wall thickness significantly improved performance: P2 had a cracking load 52.73% higher and an ultimate bearing capacity 5.7% higher than P1, with better deformation resistance and crack control. A theoretical model considering the plastic hinge mechanism at the pipe crown was developed, treating the three-edge load as an equivalent distributed plate load. The calculated results agreed well with experimental measurements. An ABAQUS finite element model successfully reproduced the full mechanical response from initial loading to failure. Parametric analysis indicated optimal performance at a hoop reinforcement ratio of approximately 1.4%. Even at 0.6%, the ultimate bearing capacity reached 367 kN/m, meeting current design code requirements. This study is novel in conducting full-scale UHPC pipe jacking tests, proposing a theoretical model accounting for crown plastic hinges, and establishing a finite element method that reproduces the entire failure process. Optimizing wall thickness and hoop reinforcement can enhance structural safety and durability, providing guidance for the design and engineering of pipe jacking structures. Full article

The manipulation of vector beams (VBs) with longitudinally variant polarization states is an important research topic and has potential applications in classical and quantum fields. In this study, we propose a half-wave plate dielectric metasurface composed of two interleaved sub-metasurfaces to generate longitudinally separated dual-channel vectorial structured light fields. The propagation and Pancharatnam–Berry phases are employed to construct hyperbolic, helical, and opposite gradient phases for focusing wavefronts, generating circularly polarized (CP) vortices, and deflecting CP vortices with the same chirality in opposite directions. Consequently, dual-channel higher-order or hybrid-order Poincaré (HOP or HyOP) beams are generated along the optical axis under elliptically polarized illumination, and their polarization states evolve along an arbitrary pair of antipodal meridians on the HOP or HyOP sphere by varying the ellipticity of the incident light, the propagation-phase topological charge, and the rotation order of the meta-atom. The consistency between the theoretical and simulated results demonstrates the feasibility and practicability of the proposed method. This study is significant for compact, integrated, and multifunctional optical devices, and provides an innovative strategy to extend optical field manipulation from two-dimensional to three-dimensional space. Full article

Fermentation is a type of biological process conducted domestically or commercially to preserve foods and beverages, produce alcohol, add nutritional value and improve aroma and flavor. The natural fermentation of flour in water to obtain a leaven for baking, lately scrutinized in the laboratory with the application of metagenomic methods, has been ubiquitous since the dawn of civilization. Commercially, single culture or defined mixtures of microorganisms are used for their predictability, but regularly fed two-domain microorganism cultures are favored in less industrialized and domestic operations. Fungi principally produce the carbon dioxide responsible for leavening. The bacteria produce acid in the bread commonly known as sourdough for its aroma and flavor. A leaven made by fermentation using flour and water can be stored while it is dormant. We studied a mature culture that is fed twenty-fold with water and flour by incubating it for 24 h, sampling it regularly for pH measurements, and plating it. The colonies were suspended for micrography and DNA extraction for PCR and Sanger sequencing. The metagenomic DNAs were analyzed for bacterial and fungal composition. The proportions of the plant and microbial DNA endogenous to the flour decline rapidly, and the predominant bacteria and fungi in mature leaven propagate, without overlap between the respective microbiomes. Full article

UK ports handle the vast majority of national trade by volume and constitute Critical National Infrastructure. Since 2004, the SOLAS/ISPS Code and the Port Security Regulations 2009 have established baseline security requirements, recently supplemented by the National Security and Investment Act 2021 and the National Security Act 2023, creating overlapping obligations. This contribution maps the evolving regulatory framework (ISPS/Port Security Regulations, NSI 2021, NSA 2023, and CNI-related guidance). It assesses operational impacts using industry metrics and draws comparative lessons from Singapore and Rotterdam. Empirical research indicates that security regulation is not uniformly detrimental to performance: targeted, intelligence-led, and technology-enabled measures can coincide with productivity gains, whereas fragmented or blanket compliance regimes are more consistently associated with increased dwell times and throughput loss. These delays propagate through supply chains and intensify cost pressures, with proportionally greater impacts on mid-sized ports. Comparative evidence indicates that risk-based screening, integrated cyber–physical platforms, transparent governance, and clear cost-sharing frameworks can maintain security without compromising commercial performance. The contribution recommends (i) tiered, risk-based screening with transparent indicators; (ii) the consolidation of overlapping regulatory obligations; (iii) clearer liability frameworks, including model terms and alternative dispute resolution; and (iv) scheduled review provisions to maintain proportionality over time. Full article

The stability of subsea tunnels is governed by the strong coupling among stress redistribution, damage evolution, and seepage flow (Stress–Damage–Seepage, SDS). The dynamic interplay, especially under high water pressure, often leads to catastrophic failures, yet its mechanisms, particularly the role of support timing, remain insufficiently understood due to limitations in conventional numerical methods. This study aims to unravel the SDS coupling mechanisms during tunnel excavation under high hydraulic head, and to quantitatively investigate how support timing influences the stability of the surrounding rock within this coupled system. A coupled Computational Fluid Dynamics and Discrete Element Method (CFD-DEM) framework was employed. In this approach, excavation-induced damage, crack propagation, and fluid–particle interactions are explicitly resolved at the particle scale, whereas the macroscopic permeability evolution is captured through an imposed empirical exponential relationship. Simulations were conducted under both steady-state and transient seepage conditions with varying stress ratios and water heads. High-head transient seepage intensifies SDS coupling, dynamically redistributing seepage forces to damage zone edges and amplifying damage. Support timing critically mediates this interaction: premature support risks tensile failure at the tunnel periphery, while delayed support allows a vicious cycle of shear failure and increased inflow. Optimal “timely” support, applied after initial deformation, diverts high seepage forces inward, minimizing final damage. The spatiotemporal synchronization of transient seepage forces with damage evolution is pivotal for stability. Support timing acts as a key control variable. The CFD-DEM framework effectively elucidates these micro-mechanisms, providing a scientific basis for the dynamic design of support in high-pressure subsea tunnels. Full article

Fault Detection and Diagnosis (FDD) in complex industrial systems requires methods that can handle uncertain operating conditions, soft thresholds, evolving sensor behavior, and increasing volumes of heterogeneous data. Traditional model-based or rule-driven approaches offer interpretability but lack adaptability, while purely data-driven and Large Language Model (LLM)-based methods often struggle with consistency, traceability, and causal grounding. Dynamic Master Logic (DML) provides a causal and temporal reasoning structure with fuzzy rules that capture gradual drift, soft limits, and asynchronous sensor signals while preserving traceability and deterministic evidence propagation. Building on this foundation, this paper presents a DML–LLM hybrid architecture that integrates targeted LLM inference to interpret unstructured information such as logs, notes, or retrieved documents under controlled prompts that maintain domain constraints. The combined system integrates Bayesian updating, deterministic routing, and semantic interpretation into a unified FDD pipeline. In a semiconductor manufacturing case study, the proposed framework reduced time to detection (TTD) from 7.4 h to 1.2 h and improved the F1 score from 0.59 to 0.83 when compared with conventional Statistical Process Control (SPC) and Fault Detection and Classification (FDC) workflows. Provenance completeness increased from 18% to 96%, while engineer triage time was reduced from 72 min to 18 min per event. These results demonstrate that the hybrid framework provides a scalable and explainable approach to anomaly detection and fault diagnosis in sensor-rich industrial environments. Full article

Atmospheric refraction often influences the localization accuracy of ground-based radar for detecting space targets. Traditional methods generally utilize the measured troposphere and ionosphere data from the local station for atmospheric refraction correction and thus neglect the influence of atmospheric horizontal inhomogeneity. However, in practice, a horizontally inhomogeneous ionosphere often causes considerable residual errors in the measured range and elevation angle after refraction correction, especially for targets with low elevation angles. The ionospheric electron density profile along the wave propagation path is significantly different from that in the vertical direction of the local station, which further brings about challenges in the modeling and correction of atmospheric refraction errors. To address the above challenge, the effect of a horizontally inhomogeneous ionosphere on the range and elevation angle measured by ground-based radar is analyzed, and a geographic division modeling strategy for the ionospheric electron density along the propagation path for atmospheric refraction correction is proposed in this paper. The simulation results show that the oblique electron density distribution obtained from the proposed model agrees well with the results calculated by the International Reference Ionosphere (IRI) model, and the proposed methodology effectively suppresses residual errors in radar atmospheric refraction correction in the low-elevation detection case. Full article

Goji, a plant unique to China, is recognized for its dual use as both a food and a medicine and is rich in various nutrients. However, long-term asexual propagation often leads to cultivar degeneration and viral accumulation, which severely impact its yield, quality, and disease resistance. Homozygous seeds can stably produce offspring with uniform traits. Haploid breeding technology, which involves doubling the chromosomes of haploid plants to obtain homozygous diploids, can significantly accelerate the breeding process. The DMP (Domain of Unknown Function 679 Membrane Protein) family is a plant-specific family of membrane proteins involved in various biological functions, including physiological processes, reproductive development, and senescence. Concurrently, loss-of-function of the DMP gene impedes the proper integration of the paternal genome following fertilization. Consequently, the embryo develops with exclusively maternal chromosomes, a mechanism that underlies the induction of haploids. In this study, we conducted a genome-wide identification of the DMP gene family in goji, analyzing the physicochemical properties, chromosomal locations, cis-acting elements, phylogenetic relationships, sequence characteristics, expression patterns, and subcellular localization of its members. The objective was to identify DMP genes capable of inducing haploid production in goji berry for future breeding applications. The results revealed a total of 11 DMP family members in the goji berry genome, distributed across seven chromosomes. The proteins encoded by these members contain 136 to 237 amino acids, with molecular weights ranging from 15,267.96 to 26,141.01 Da and isoelectric points (pI) ranging from 5.14 to 9.32. The LbDMPs were found to contain numerous cis-acting elements that play roles in plant responses to abiotic stresses and various phytohormones. Notably, LbDMP1 and LbDMP11, which contain the typical DUF679 domain, are predominantly expressed in pollen, suggesting their involvement in the reproductive process of goji berry. They were therefore identified as candidate genes for haploid induction. Subcellular localization analysis demonstrated that LbDMP1 is localized to the plasma membrane, while LbDMP11 is localized to membrane systems such as the endoplasmic reticulum. This research provides a fundamental basis for further exploration of the functional roles of the DMP gene family in goji berry and offers valuable genetic resources for haploid induction in its breeding programs. Full article

Grain-scale heterogeneity strongly influences hydraulic fracture initiation and trajectory in crystalline rocks, yet its contributions to non-planar growth and the interaction of multiple nearby cracks remain insufficiently quantified. To address this gap, we perform numerical experiments on a model containing two parallel pre-existing cracks using a hydro-mechanical phase-field framework, systematically quantifying how mineral distribution and axial compression govern non-planar hydraulic fracture growth and inter-fracture competition. The results demonstrate that mineral distribution is the primary driver of fracture complexity. Even within the same Voronoi tessellation, redistributing minerals alone yields markedly different trajectories, deflections, branching patterns, and final morphologies. Furthermore, non-planar growth follows a stepwise, energy-threshold-driven mechanism. When cracks penetrate strong grains or undergo large-angle deflections, propagation is impeded, and injection pressure builds up. Once a critical energy threshold is reached, accumulated energy is rapidly released along the path of minimum incremental energy, manifested as abrupt pressure drops and rapid crack advance. Additionally, the two nearby fractures exhibit strong mechanical competition. Despite negligible hydraulic interference in low-permeability granite, early growth of one fracture redistributes stresses and suppresses the driving force of the other, resulting in asymmetric development. Finally, axial compression primarily governs the overall propagation orientation and influences local failure modes but has a limited effect on peak pressure relative to mineral distribution. Full article

With the increasing deployment of autonomous driving systems, the opaque nature of deep reinforcement learning (DRL) decision models hinders understanding and validation of driving decisions. To address this challenge, we propose a Hybrid Attribution-based Interpretable Deep Reinforcement Learning framework (HA-IDRL) for autonomous driving behavior decision-making. The framework introduces a Hybrid Gradient–LRP (HGL) attribution mechanism that integrates gradient-based attribution and Layer-wise Relevance Propagation (LRP) to capture complementary sensitivity and contribution information, producing more consistent and comprehensive post hoc explanations. In addition to post hoc interpretability, we enhance structural interpretability by replacing the conventional multilayer perceptron (MLP) in the Dueling Deep Q-Network (Dueling DQN) architecture with Kolmogorov–Arnold Networks (KAN). By representing nonlinear interactions through learnable univariate functions and explicit summation structures, KAN provides inherently interpretable functional decompositions. The proposed framework is evaluated on a highway lane-changing task using the highway-env simulator. Experimental results show that HA-IDRL achieves decision-making performance comparable to representative DRL baselines, including Dueling DQN and Soft Actor-Critic (SAC), while providing explanations that are more stable and better aligned with human driving semantics. Moreover, the proposed method produces explanations with low computational overhead, enabling efficient and real-time interpretability in practical autonomous driving applications. Overall, HA-IDRL advances trustworthy autonomous driving by enabling high-performance decision-making and rigorous, multi-level interpretability, thereby improving the transparency and operational reliability of DRL-based driving policies. Full article

The ice–rock interface shear mechanism is fundamental to understanding ice–rock avalanche hazards. This study conducts a series of direct shear tests under various normal stresses to analyze the mechanical response and acoustic emission (AE) evolution of the interface, establishing a shear strength prediction model. Results indicate that the roughness significantly affects mechanical properties and AE responses: as the roughness increases, the shear strength, cohesion, and internal friction angle improve significantly, while peak AE ringing counts and energy exhibit an increasing trend. During failure, the proportion of shear cracks decreases while tensile cracks increase, reflecting a shift in crack development modes driven by the roughness. Based on AE characteristics and stress–displacement relations, the shear failure process is categorized into five stages: initial, crack development, crack propagation, crack coalescence, and residual stages. Incorporating the effects of the roughness and cementation force, a shear mechanical model was established. Experimental data verify the model’s rationality; however, its applicability may be limited when the roughness is excessively high. Full article

Waveguide-like propagation in elongated underground environments—utility corridors, logistics tunnels—generates dense multipath that can cause the earliest or strongest resolvable channel impulse response (CIR) component to originate from a specular reflection rather than the direct line-of-sight (LOS) path. In the single-anchor CIR-tap-based implementations common to practical ultra-wideband (UWB) systems, baseline estimators such as phase-difference-of-arrival (PDOA) and MUSIC rely on selecting a single dominant CIR component, producing large angle-of-arrival (AoA) errors whenever the selected path is a reflection. We propose a multipath credibility selection (MCS) AoA estimator, MCS-AoA, that does not require explicit LOS/NLOS classification. The algorithm scores each resolvable CIR component with four credibility factors—amplitude significance, time-of-flight (TOF) consistency, inter-baseline phase–geometry agreement, and cross-baseline coherence—and fuses retained candidates into a credibility-weighted spatial covariance matrix for 2D MUSIC search. Field experiments on a custom five-channel coherent UWB platform compare MCS-AoA against six baselines—PDOA, MUSIC, MVDR/Capon, TLS-ESPRIT, PwMUSIC, and DNN-AoA. In an underground corridor (5–40 m), MCS-AoA achieves an azimuth/elevation MAE of 1.00°/1.46°, outperforming all baselines (PDOA: 2.26°/2.49°; MUSIC: 1.76°/2.40°; next-best PwMUSIC: 1.44°/2.17°); in a logistics tunnel (5–80 m), it achieves a 1.19° overall azimuth MAE. Simulations corroborate these gains, with a 0.71° azimuth RMSE at 80 m (69.3% reduction over PDOA) and 86.6% of estimates falling within 1°. Full article