Search Results (283)

Objective: We examine whether a finite-range scalar–tensor modification of gravity can be simultaneously compatible with cosmological background data, galaxy rotation curves, and local/astrophysical consistency tests, while satisfying the luminal gravitational-wave propagation constraint ($c_T=1$) implied by GW170817 at low redshifts. Methods: We formulate the model at the level of an explicit covariant action and derive the corresponding field equations; for cosmological inferences, we adopt an effective background closure in which the late-time dark-energy density is modulated by a smooth activation function characterized by a length scale $\lambda$ and amplitude $ϵ$. We constrain this background model using Pantheon+, DESI Gaussian Baryon Acoustic Oscillations (BAOs), and a Planck acoustic-scale prior, including an explicit $\Lambda$CDM comparison. We then propagate the inferred characteristic length by fixing $\lambda$ in the weak-field Yukawa kernel used to model 175 SPARC galaxy rotation curves with standard baryonic components and a controlled spherical approximation for the scalar response. Results: The joint background fit yields ${\Omega }_m=0.293\pm 0.007$, $\lambda =7.{69}_{-1.71}^{+1.85}$$\mathrm{Mpc}$, and $H_0=72.33\pm 0.50$$\mathrm{km}{\mathrm{s}}^{-1}{\mathrm{Mpc}}^{-1}$. With $\lambda$ fixed, the baryons + scalar model describes the SPARC sample with a median reduced chi-square of ${\chi }_{\nu }^2=1.07$; for a 14-galaxy subset, this model is moderately preferred over the standard baryons + NFW halo description in the finite-sample information criteria, with a mean $\Delta$AICc outcome in favor of the baryons + scalar model (≈2.8). A Vainshtein-type screening completion with $\Lambda =1.3\times {10}^{-8}$ eV satisfies Cassini, Lunar Laser Ranging, and binary pulsar bounds while keeping the kpc scales effectively unscreened. For linear growth observables, we adopt a conservative General Relativity-like baseline (${\mu }_0=0$) and show that current $f{\sigma }_8$ data are consistent with ${\mu }_0\simeq 0$ for our best-fit background; the model predicts $S_8=0.791$, consistent with representative cosmic-shear constraints. Conclusions: Within the present scope (action-level weak-field dynamics for galaxy modeling plus an explicitly stated effective closure for background inference), the results support a mutually compatible characteristic length at the Mpc scale; however, a full perturbation-level implementation of the covariant theory remains an issue for future work, and the role of cold dark matter beyond galaxy scales is not ruled out. Full article

Multi-key homomorphic encryption (MKHE) is crucial for secure collaborative computing, yet it suffers from high multiplicative depth and computational overhead during Look-Up Table (LUT) evaluations, particularly for large input domains. To address these challenges, this paper proposes an optimized LUT evaluation method based on multivariate polynomial approximation. Specifically, we partition the high-dimensional input space into several lower-dimensional variables to design low-depth multivariate polynomials. By integrating blockwise encoding and tensor-based transformations, we construct a parallelizable evaluation framework that maps multivariate functions into a high-dimensional polynomial-coefficient space. This approach allows for efficient parallel processing and effective noise management. Theoretical analysis demonstrates that our method significantly reduces the multiplicative depth from O(ℓ) to O(ℓ/α), indicating its robustness and efficiency in large-scale LUT scenarios. Full article

Neural network quantization has become established as a key strategy for transitioning medical imaging models from research environments to clinical devices and resource-constrained edge platforms; however, the available evidence remains fragmented and focused on highly heterogeneous use cases. This study presents a systematic review of 72 studies on quantization applied to medical images, following PRISMA guidelines, with the aim of characterizing the relationship among quantization technique, network architecture, imaging modality, and execution environment, as well as their impact on latency, memory footprint, and clinical deployment. Based on a structured variable matrix, we analyze—through tailored visualizations—usage patterns of Post-Training Quantization (PTQ), Quantization-Aware Training (QAT), mixed precision, and binary/low-bit schemes across frameworks such as PyTorch V 2.6.0, TensorFlow 2.19.0, and TensorFlow Lite, executed on server-class GPUs, edge/embedded devices, and specialized hardware. The results reveal a strong concentration of evidence in PyTorch/TensorFlow pipelines using INT8 or mixed precision on GPUs and edge platforms, contrasted with limited attention to PACS/RIS interoperability, model lifecycle management, energy consumption, cost, and regulatory traceability. We conclude that, although quantization can approximate real-time performance and reduce memory footprint, its clinical adoption remains constrained by integration challenges, model governance requirements, and the maturity of the hardware–software ecosystem. Full article

In this paper, an anisotropic object electromagnetic image reconstruction system based on a two-stage multi-frequency extended network is developed by deep learning techniques. We obtain the scattered field information by irradiating the TM different polarization waves to uniaxial objects located in free space. We input the measured single-frequency scattered field into the Deep Residual Convolutional Neural Network (DRCNN) for training and to be further extended to multi-frequency data by the trained model. In the second stage, we feed the multi-frequency data into the Deep Convolutional Encoder–Decoder (DCED) architecture to reconstruct an accurate distribution of the dielectric constants. We focus on EMIS applications using Transverse Magnetic (TM) and Transverse Electric (TE) waves in 2D scenes. Numerical findings confirm that our method can effectively reconstruct high-contrast uniaxial objects under limited information. In addition, the TM/TE scattering from uniaxial anisotropic objects is governed by polarization-dependent Lippmann–Schwinger integral equations, yielding a nonlinear and severely ill-posed inverse operator that couples the dielectric tensor components with multi-frequency field responses. Within this mathematical framework, the proposed two-stage DRCNN–DCED architecture serves as a data-driven approximation to the anisotropic inverse scattering operator, providing improved stability and representational fidelity under limited-aperture measurement constraints. Full article

Background: Approximately 20–30% of ultra-high risk (UHR) individuals transition to psychosis within 2–3 years. Neurobiological markers predicting conversion remain critical for precision prevention strategies. Objective: To systematically identify and evaluate structural and functional neuroimaging biomarkers at UHR baseline that predict subsequent conversion to psychosis. Methods: Following PRISMA 2020 guidelines, we searched five databases from January 2000 to February 2025. Two independent reviewers screened studies and assessed quality using the Newcastle–Ottawa Scale. Eligible studies examined baseline neuroimaging measures (structural MRI, functional MRI, diffusion tensor imaging, magnetic resonance spectroscopy) as predictors of psychosis conversion in UHR cohorts. Results: Twenty-five studies comprising 2436 UHR individuals (627 converters, 25.7%) were included (80.0% high quality). Reduced baseline gray matter volume in medial temporal structures (hippocampus: Cohen’s d = −0.45 to −0.68; parahippocampal gyrus: d = −0.52 to −0.71) and prefrontal cortex (d = −0.41 to −0.68) consistently predicted conversion. Progressive gray matter loss in superior temporal gyrus distinguished converters (d = −0.72). Reduced prefrontal–temporal functional connectivity predicted conversion (AUC = 0.73–0.82). Compromised white matter integrity in uncinate fasciculus (fractional anisotropy: d = −0.47 to −0.71) and superior longitudinal fasciculus predicted transition. Elevated striatal glutamate predicted conversion (d = 0.52–0.76). Thalamocortical dysconnectivity showed large effects (Hedges’ g = 0.66–0.88). Multimodal imaging models achieved 78–85% classification accuracy. Conclusions: Neuroimaging biomarkers, particularly medial temporal and prefrontal structural alterations, functional dysconnectivity, and white matter abnormalities, demonstrate moderate-to-large effect sizes in predicting UHR conversion. Multimodal approaches combining structural, functional, and neurochemical measures show promise for individualized risk prediction and early intervention targeting in precision prevention strategies. Full article

In this contribution, we propose and study long-time behaviors of a new class of N-dimensional delayed Kirchhoff–Carrier-type problems with variable transfer coefficients involving a logarithmic nonlinearity. We take into account the dependence of diffusion and damping coefficients on the position and direction, as well as the presence of different types of delays. This class of nonlocal anisotropic and nonlinear wave-type equations with multiple time-varying mixed delays and dampings, of a fairly general form, containing several arbitrary functions and free parameters, is of the following form: ${\displaystyle \frac{{\partial }^{2}u}{\partial t^2}-{div(\mathcal{K}(\Vert \sigma \nabla u\Vert }_{L^2(\Omega )}^2)A_{\sigma }{(\mathbf{x})\nabla u)+\mathcal{M}(\Vert u\Vert }_{L^2(\Omega )}^2)u-div(\zeta (t)A_{\sigma }(\mathbf{x})\nabla \frac{\partial u}{\partial t})+d_0(t)\frac{\partial u}{\partial t}+{\mathbb{D}}_r(\mathbf{x},t;\frac{\partial u}{\partial t})=\mathbb{G}(u),}$ where $u(\mathbf{x},t)$ is the state function, $\mathcal{M}$ and $\mathcal{K}$ are the nonlocal Kirchhoff operators and the nonlinear operator $\mathbb{G}(u)$ corresponds to a logarithmic source term. The symmetric tensor $A_{\sigma }$ describes the anisotropic behavior and processes of the system, and the operator ${\mathbb{D}}_r$ represents the multiple time-varying mixed delays related to velocity ${\displaystyle \frac{\partial u}{\partial t}}$. Our problem, which encompasses numerous equations already studied in the literature, is relevant to a wide range of practical and concrete applications. It not only considers anisotropy in diffusion, but it also assumes that the strong damping can be totally anisotropic (a phenomenon that has received very little mathematical attention in the literature). We begin with the reformulation of the problem into a nonlinear system coupling a nonlocal wave-type equation with ordinary differential equations, with the help of auxiliary functions. Afterward, we study the local existence and some necessary regularity results of the solutions by using the Faedo–Galerkin approximation, combining some energy estimates and the logarithmic Sobolev inequality. Next, by virtue of the potential well method combined with the Nehari manifold, conditions for global in-time existence are given. Finally, subject to certain conditions, the exponential decay of global solutions is established by applying a perturbed energy method. Many of the obtained results can be extended to the case of other nonlinear source terms. Full article

Network intrusion detection is a key technology for ensuring cybersecurity. However, current methods face two major challenges: reliance on manual feature engineering, which leads to the loss of discriminative information, and the vulnerability of deep learning models to adversarial sample attacks. To address these issues, this paper proposes STS-AT, a novel network intrusion detection method that integrates structured tensors with adversarial training. The method consists of three core components: first, structured tensor encoding, which fully converts raw hexadecimal traffic into a numerical representation; second, a hierarchical deep learning model that combines CNN and LSTM networks to simultaneously learn spatial and temporal features of the traffic; third, a multi-strategy adversarial training method that enhances model robustness by adaptively adjusting the mix of adversarial samples in different training phases. Experiments on the CICIDS2017 dataset show that the proposed method achieves an accuracy of 99.6% in normal traffic classification, significantly outperforming classical machine learning baselines such as Random Forest (93.1%) and Support Vector Machine (84.7%). Crucially, under various adversarial attacks (FGSM, PGD, and DeepFool), the accuracy of an undefended model drops to as low as 24.4%, whereas after multi-strategy adversarial training, the defense accuracy rises above 96.8%. Meanwhile, the total training time is reduced by approximately 67.6%. These results verify that structured tensor encoding effectively preserves original traffic information, the hierarchical model achieves comprehensive feature learning, and multi-strategy adversarial training significantly improves training efficiency while ensuring robust defense effectiveness. Full article

This research article focuses on the method of vehicle crash velocity evaluation based on the tensor product approximation by B-splines. Weights associated with each observation are introduced in the least square method, which is based on probabilistic reasoning. The presented calculation algorithm is built based on the Compact vehicle class from the NHTSA database, which consists of 338 records of frontal crashes with stationary obstacles. The presented model is restricted to velocities in the range between 43 and 93.5 km/h. The relative error obtained for the presented calculation method is 5.2%. Hence, improvement has been observed in comparison with the linear model with the same probabilistic weights approach, for which relative error is equal to 5.47%. Full article

This paper presents a unified framework for constructing partially unstructured B-spline transfinite finite elements with arbitrary nodal distributions. Three novel, distinct classes of elements are investigated and compared with older single Coons-patch elements. The first consists of classical transfinite elements reformulated using B-spline basis functions. The second includes elements defined by arbitrary control point networks arranged in parallel layers along one direction. The third features arbitrarily placed boundary nodes combined with a tensor-product structure in the interior. For all three classes, novel macro-element formulations are introduced, enabling flexible and customizable nodal configurations while preserving the partition of unity property. The key innovation lies in reinterpreting the generalized coefficients as discrete samples of an underlying continuous univariate function, which is independently approximated at each station in the transfinite element. This perspective generalizes the classical transfinite interpolation by allowing both the blending functions and the univariate trial functions to be defined using non-cardinal bases such as Bernstein polynomials or B-splines, offering enhanced adaptability for complex geometries and nonuniform node layouts. Full article

To address the limitations of traditional non-local mean (NLM) denoising algorithms in terms of neighborhood similarity metrics, weight calculation, and computational efficiency, this paper proposed a structural-tensor-driven and dynamic window-based fast non-local mean denoising algorithm with dual kernel weighting. First, a Gaussian–Tukey dual-kernel weighting function was designed to optimize similarity metrics. Then, spatial neighborhood features were adopted. By measuring both grayscale similarity and spatial correlation, the weight distribution rationality was further enhanced. Second, structural tensor eigenvalues were used to quantify regional structural properties. A dynamic window allocation function was designed to adaptively match search window sizes to different image regions. Finally, an integral image acceleration mechanism was proposed, significantly improving algorithm execution efficiency. Experimental results demonstrated that the proposed algorithm achieved both excellent denoising performance and edge/texture preservation capabilities. In high-noise environments, its Peak Signal-to-Noise Ratio (PSNR) outperformed the Gauss kernel non-local mean algorithm by an average of 1.96 dB, while Structural Similarity (SSIM) improved by an average of 5.7%. Moreover, the algorithm’s execution efficiency increased by approximately 7–11 times, indicating strong potential for real-time application in digital image processing. Full article

Numerical simulation of groundwater level dynamics plays a crucial role in the safety assessment of near-surface radioactive waste disposal facilities. Such disposal sites are typically located in regions characterized by extensive bedrock outcrops. However, accurately characterizing the permeability of fractured media is challenging, and the scarcity of groundwater level data poses significant difficulties for constructing reliable numerical models. This study focuses on a near-surface disposal site in northwestern China. By integrating field packer tests with hydraulic conductivity tensors computed from borehole televiewer data, we quantitatively evaluated the permeability of fractured rocks of different lithologies to provide accurate parameters for numerical modeling. The constructed groundwater flow model was further calibrated and validated using long-term groundwater level monitoring data and field tracer-based groundwater flow direction tests, ensuring high model reliability. Using the calibrated model, groundwater level variations were simulated under various rainfall and pumping scenarios. The results show that pumping intensity in the downstream farmland area exerts a limited influence on groundwater levels beneath the disposal site, while rainfall intensity plays a dominant role. Under the heavy rainfall scenario, the groundwater level at the disposal site rises by approximately 5.2 m after 50 years, leaving a 6 m gap above the base of the disposal unit. Under prolonged heavy rainfall conditions, implementing drainage measures may be necessary to ensure the repository’s long-term safety. Full article

We suggest a novel strategy in the theory of elastic plane composites. The macroscopic properties are quantified, and an analytical–numerical algorithm to derive expressions for the effective constants is designed. The effective elastic constants of dispersed random composites are given by new analytical and approximate formulas where the dependence on the location of inclusions is explicitly shown in symbolic form. This essentially extends the results of previous numerical simulations for a fixed set of material constants and fixed locations of inclusions. This paper extends the analysis from Part I, which addressed dispersed random conducting composites, to the two-dimensional elastic composites. Hill’s concept of Representative Volume Element (RVE), traditionally used in elastic composites, is revised. It is rigorously demonstrated that the RVE must be a fundamental domain of the plane torus, for instance, a periodicity parallelogram, since other shapes of RVE may lead to incorrect values of the effective constants. The effective tensors of the elasticity theory are decomposed into geometrical and physical parts, represented by structural sums and material constants of the components. Novel computational methodology based on such decomposition is applied to a two-phase isotropic composite with non-overlapping circular inclusions embedded in an elastic matrix. For the first time, it is demonstrated explicitly how the effective tensors depend on the geometric probabilistic distributions of inclusions and the computational protocols involved. Analytical polynomial formulas for the effective shear modulus for the moderate concentration of inclusions are transformed using the resummation methods into practical expressions valid for all concentrations of inclusions. The critical index for the effective shear modulus is calculated from the polynomials derived for the modulus. Full article

Maternal health care during labor requires the continuous and reliable monitoring of analgesic procedures, yet conventional systems are often subjective, indirect, and operator-dependent. Infrared thermography (IRT) offers a promising non-invasive approach for labor epidural analgesia (LEA) monitoring, but its practical implementation is hindered by clinical and hardware limitations. This work presents a novel artificial intelligence-driven mobile platform to overcome these hurdles. The proposed solution integrates a lightweight deep learning model for semantic segmentation, a B-spline-based free-form deformation (FFD) approach for non-rigid dermatome registration, and efficient on-device inference. Our analysis identified a U-Net with a MobileNetV3 backbone as the optimal architecture, achieving a high Dice score of 0.97 and a 4.5% intersection over union (IoU) gain over heavier backbones while being 73% more parameter-efficient. The entire AI pipeline is deployed on a commercial smartphone via TensorFlow Lite, achieving an on-device inference time of approximately two seconds per image. Deployed within a user-friendly interface, our approach provides straightforward feedback to support decision making in labor management. By integrating thermal imaging with deep learning and mobile deployment, the proposed system provides a practical solution to enhance maternal care. By offering a quantitative, automated tool, this work demonstrates a viable pathway to augment or replace subjective clinical assessments with objective, data-driven monitoring, bridging the gap between advanced AI research and point-of-care practice in obstetric anesthesia. Full article

The increasing need for high-resolution, real-time radiative transfer (RT) modeling in climate science, remote sensing, and planetary exploration has exposed limitations of traditional solvers such as the Discrete Ordinate Radiative Transfer (DISORT) and Rapid Radiative Transfer Model for General Circulation Models (RRTMG), particularly in handling spectral complexity, non-local thermodynamic equilibrium (non-LTE) conditions, and computational scalability. Quantum-Inspired Neural Radiative Transfer (QINRT) frameworks, combining tensor-network parameterizations and quantum neural operators (QNOs), offer efficient approximation of high-dimensional radiative fields while preserving key physical correlations. This review highlights the advances of QINRT in enhancing spectral fidelity and computational efficiency, enabling energy-efficient, real-time RT inference suitable for satellite constellations and unmanned aerial vehicle (UAV) platforms. By integrating physics-informed modeling with scalable neural architectures, QINRT represents a transformative approach for next-generation Earth-system digital twins and autonomous climate intelligence. Full article

This paper develops mathematical apparatus for the modeling of the electronic structure of semiconductor nanoparticles and the description of sensor response of the layers constructed on their base. The developed technique involves solutions of both the direct and inverse problems. The direct problem involves of the two coupled sets of differential equations, at fixed values of physical parameters. The first of them is the set of equations of chemical kinetics which describes processes occurring at the surface of a nanoparticle. The second involves an equation describing electron concentration distribution inside a nanoparticle. The inverse problem consists of the determination of physical parameters (essentially, reactions rate constants) which provide a good approximation of experimental data when using them to find the solution of the direct problem. The mathematical novelty of this paper is the application of—for the first time, to find the solution of the inverse problem—the new gradient-free optimization methods based on low-rank tensor train decomposition and modern machine learning paradigm. Sensor effect was measured in a dedicated set of experiments. Comparisons of computed and experimental data on sensor effect were carried out and demonstrated sufficiently good agreement. Full article