Search Results (4,022)

The demand for high-quality private data in large language models is growing significantly. However, private data is often scattered across different entities, leading to significant data silo issues. To alleviate such problems, we propose a novel multi-party collaborative training framework for large language models, named MPCTF. MPCTF consists of several components to achieve multi-party collaborative training: (1) a one-click launch mechanism with multi-node and multi-GPU training capabilities, significantly simplifying user operations while enhancing automation and optimizing the collaborative training workflow; (2) four data partitioning strategies for splitting client datasets during the training process, namely fixed-size strategy, percentage-based strategy, maximum data volume strategy, and total data volume and available GPU memory strategy; (3) multiple aggregation strategies; and (4) multiple privacy protection strategies to achieve privacy protection. We conducted extensive experiments to validate the effectiveness of the proposed MPCTF. The experimental results demonstrate that the proposed MPCTF achieves superior performance; for example, our MPCTF acquired an accuracy rate of 65.43 and outperformed the existing work, which acquired an accuracy rate of 14.25 in the experiments. Moreover, we hope that our proposed MPCTF can promote the development of collaborative training for large language models. Full article

The aim of this study was to address the issue of significant performance degradation in existing defogging algorithms under extreme fog conditions. Traditional Taylor series-based deformable convolutions are limited by local approximation errors, while the heavy-tailed characteristics of the Cauchy distribution can more successfully model outliers in fog images. The following improvements are made: (1) A displacement generator based on the inverse cumulative distribution function (ICDF) of the Cauchy distribution is designed to transform uniform noise into sampling points with a long-tailed distribution. A novel double-peak Cauchy ICDF is proposed to dynamically balance the heavy-tailed characteristics of the Cauchy ICDF, enhancing the modeling capability for sudden changes in fog concentration. (2) An innovative Cauchy–Gaussian fusion module is proposed to dynamically learn and generate hybrid coefficients, combining the complementary advantages of the two distributions to dynamically balance the representation of smooth regions and edge details. (3) Tree-based multi-path and cross-resolution feature aggregation is introduced, achieving local–global feature adaptive fusion through adjustable window sizes (3/5/7/11) for parallel paths. Experiments on the RESIDE dataset demonstrate that the proposed method achieves a 2.26 dB improvement in the peak signal-to-noise ratio compared to that obtained with the TaylorV2 expansion attention mechanism, with an improvement of 0.88 dB in heavily hazy regions (fog concentration > 0.8). Ablation studies validate the effectiveness of Cauchy distribution convolution in handling dense fog and conventional lighting conditions. This study provides a new theoretical perspective for modeling in computer vision tasks, introducing a novel attention mechanism and multi-path encoding approach. Full article

The decline and revitalization of vitality in historic districts of small- and medium-sized cities undergoing rapid urbanization is a frontier issue in global heritage conservation and urban regeneration. Using the East Street Historic District in Mengzi, Yunnan, as a case study, this study proposes a “space–function–time” coupling framework. Topological accessibility is quantified through space syntax metrics—Integration Value (2021) and Integration Value (2025), as well as Choice Value (2021) and Choice Value (2025)—while functional aggregation is represented by POI kernel density analysis. A “Deviation Degree–Change in Deviation Degree” model is developed to track the dynamic evolution before and after the implementation of the conservation plan (2021–2025). The findings indicate that (1) the linear correlation between Integration Value and POI density decreases from a moderate level (r = 0.42) in 2021 to a weak correlation (r = 0.32) in 2025, revealing that the spatial–functional coordination mechanism in small- and medium-sized city historic districts is considerably more fragile than in large cities; (2) Identifying streets with abnormal deviations: The primary street, Renmin Middle Road, exhibits a deviation degree as high as 4.160 due to excessive commercial aggregation, resulting in a “high accessibility–high load” imbalance. The secondary street, Dashu Street, although demonstrating a relatively high Integration Value (0.663), shows a “high accessibility–low vitality” condition due to insufficient functional facilities; (3) the Deviation Degree–Change in Deviation Degree model accurately identifies High Deviation Streets, Medium Deviation Streets, and Low Deviation Streets, and provides quantitative thresholds for planning feedback. This study introduces the Deviation Degree–Change in Deviation Degree model for the first time into the evaluation of historic district renewal in small- and medium-sized cities, establishing a closed-loop “diagnosis–intervention–reassessment” tool. The proposed framework offers both a methodological and operational paradigm for precision-oriented urban regeneration in historic districts. Full article

The Nahuelbuta Mountain Range in the south-central zone of Chile is a biodiversity hotspot that has undergone intense land use/cover transformation. This study analyses two decades of land use change (1999–2018) in the watersheds of the Lebu and Leiva rivers. The magnitude and spatial configuration of these changes were assessed using detailed spatial information, transition matrices, and landscape metrics. The results show that between 1999 and 2018, temperate native forest decreased by 30.3% in Lebu and 22.8% in Leiva, being replaced mainly by exotic forest plantations, which increased by 20.2% and 13.5%, respectively. The spatiotemporal analysis revealed losses concentrated in the lowland and middle zones of both watersheds, with persistence of temperate native forest in higher elevations. Landscape metrics showed an increase in diversity (SIDI: Lebu 0.41–0.65; Leiva 0.29–0.57) and a decrease in aggregation (AI: 92–86%; 95–90%). At the class level, the temperate native forest presented greater fragmentation, with a reduction in size and proximity, an increase in density, and more irregular shapes. In contrast, exotic forest plantations increased in size and proximity, with a slight decrease in density and greater complexity of form, consolidating their expansion and spatial continuity in both watersheds. These findings emphasise the need to implement territorial planning and conservation strategies adapted to the Nahuelbuta context, through native forest management plans that promote ecological conservation, the recovery of degraded landscapes, and the strengthening of ecosystem services, thus contributing to the well-being of local communities and long-term environmental sustainability. Full article

The Qinghai–Tibet Plateau presents a unique challenge for infrastructure development due to its extreme geological and climatic conditions—high elevation, large diurnal temperature fluctuations, frequent freeze–thaw cycles, intense ultraviolet radiation, and seasonal precipitation. These factors greatly accelerate the weathering of rock materials, leading to aggregates with increased porosity, microcracking, and weakened mechanical properties. While the engineering implications of such degradation are evident, the underlying material science of weathered aggregates—particularly their microstructure–property relationships—remains insufficiently explored, necessitating further investigation to inform material selection and design. In this study, three representative types of weathered aggregates (silica-rich, carbonaceous, and alumina-rich), alongside unweathered natural aggregates, were examined through both macro-scale (density, water absorption, crushing value, abrasion resistance) and micro-scale (scanning electron microscopy (SEM), energy-dispersive spectroscopy (EDS)) analyses. To capture the material evolution, we introduced a simplified classification framework based on the Si/Al ratio and porosity and applied a gray entropy correlation model to quantify the coupling between microstructure and mechanical performance. Results show that weathering reduces the Si/Al ratio from 2.45 to 1.82, increases porosity from 4.2% to 12.7%, enlarges the average pore size to 0.85 μm, raises microcrack density to 1.40 μm/μm², and increases the proportion of connected pores to 68.2%. These microstructural degradations correlate with decreased aggregate density, increased water absorption (up to 8.0%), higher crushing value (27.4%), and abrasion resistance loss (26.0%). Based on these findings, a weathered aggregate classification and pretreatment strategy is proposed, offering a practical reference for engineers to improve material performance in high-altitude road construction. Full article

Urban waterlogging due to sudden rainfall leads to critical issues. This study aims to develop sustainable porous asphalt pavement by incorporating different proportions of recycled coarse aggregate. Recycled coarse aggregate from waste laboratory-tested concrete in 19, 12.5, and 9.5 mm sizes was prepared for a porous asphalt mix series (Mix-Types 1-9). The study showed that optimal aggregate ratios performed well in porous asphalt mixes. Mix-Type-3 with the aggregate ratio of 19:12.5:9.5 mm (1:1:0.5) achieved an optimal stability of 8.88 kN at the minimum flow rate. The movement of water flow results revealed that permeability decreases with time. The Mix-Type-3 permeability reductions were found to be 16.75% and 30.14% at 6 and 12 months, compared to the permeability of fresh mixes. The study results revealed that the Mix-Type-3 retained the highest stability level, and the permeability fell within the standard values. Hence, it is concluded that Mix-Type-3 balances in all parameters and is a viable choice for effective and sustainable urban water management. Full article

Amino acids are not just monomers of proteins, but they can also carry biological functions. L-cysteine (Cys), L-proline (Pro), L-asparagine (Asn), and L-glutamic acid (Glu) were used to evaluate how different amino acid chemistries alter the morphology and size of the silver nanoparticles (AgNPs) synthesized in the presence of two carbohydrate ligands, which were lactose methoxyaniline (LMA) and galactose 5-aminosalicylic acid (G5AS). UV–vis, infrared (IR), High-Resolution Transmission Electron Microscopy (HR-TEM) and X-ray diffraction (XRD) characterizations revealed that the effect of amino acids on the characteristics of the AgNPs showed dependence on the carbohydrate ligand chemistry. In the case of LMA, AgNPs shifted from aggregates to anisotropic nanoparticles, larger aggregates, and a mixture of anisotropic and 1D nanoparticles in the presence of Cys, Glu, Asn and Pro amino acids, respectively. In contrast to this, the introduction of Cys and Asn caused the formation of cluster-like AgNPs and larger rounded nanoparticles, while G5AS-synthesized AgNPs were multigonal 0D particles. Moreover, Glu and Pro contributed the resistance of silver oxide formation on the particles. Antibacterial characterization showed that LMA_Glu_AgNPs were the most effective ones, while LMA_Cys_AgNPs and G5AS_Cys_AgNPs, which were the smallest AgNPs, did not show any significant antibacterial activity. Full article

Three-dimensional reconstruction programs are essential tools for understanding the behavior of asphalt mixtures. On the basis of accurate 3D models, it is convenient to identify the complex relationship between spatial structures and physical properties. In this work, we explore a low-cost and data-efficient way to create a collection of 3D asphalt mixture models. The core idea is to introduce a foundational segmentation program and stochastic modeling into the asphalt mixture reconstruction framework. First, our approach captures a 2D image to present spatial structures of the investigated sample. The integration of a smartphone camera and an image quilting method has been designed to understand fine-grained details and facilitate full coverage. Aiming at realizing high-quality segmentation, we propose the Segment Anything Model (SAM)-driven method to distinguish aggregate grains and asphalt binder. Second, Multiple-Point Statistics (MPS) is activated to build 3D models from 2D training images. To speed up the reconstruction step, we apply Nearest Neighbor Simulation (NNSIM) to improve pattern searching efficiency. Aiming at calculating 3D conditional probabilities, the probability aggregation framework is introduced into the asphalt mixture investigation. Third, our program focuses on the modeling evaluation procedure. Determination of a two-point correlation function, analysis of distance and a grain size distribution assessment are separately performed to check the reconstruction quality. The evaluation results indicate that our program not only preserves spatial patterns but also expresses uncertainty during the material production step. Full article

The overexpression of atypical protein kinase C-iota (PKC-ι) is a biomarker for carcinogenesis in various cell types, such as glioma, ovarian, renal, etc., manifesting as a potential drug target. In previous in vitro studies, ICA-1S and ICA-1T, experimental candidates for inhibiting PKC-ι, have demonstrated their specificity and promising efficacy against various cancers. Moreover, the in vivo studies have demonstrated low toxicity levels in acute and chronic murine models. Despite these prior developments, the binding affinities of the inhibitors were never thoroughly explored from a biophysical perspective. Here, we present the biophysical characterizations of PKC-ι in combination with ICA-1S/1T. Various methods based on molecular docking, light scattering, intrinsic fluorescence, thermal denaturation, and heat exchange were applied. The biophysical characteristics including particle sizing, thermal unfolding, aggregation profiles, enthalpy, entropy, free energy changes, and binding affinity (K_d) of the PKC-ι in the presence of ICA-1S were observed. The studies indicate the presence of domain-specific stabilities in the protein–ligand complex. Moreover, the results indicate a spontaneous reaction with an entropic gain, resulting in a possible entropy-driven hydrophobic interaction and hydrogen bonds in the binding pocket. Altogether, these biophysical studies reveal important insights into the binding interactions of PKC-ι and its inhibitors ICA-1S/1T. Full article

Solvent-exchange-induced in situ forming gel (ISG) refers to a drug delivery system that transforms from a solution state into a gel or solid matrix upon administration into the body and exposure to physiological aqueous fluid. This study investigates the molecular behavior and phase inversion process of cellulose acetate butyrate (CAB)-based in situ forming gel (ISG) formulations containing moxifloxacin (Mx) or benzydamine HCl (Bz) as model drugs dissolved in N-methyl pyrrolidone (NMP) using molecular dynamics (MD) simulations and density functional theory (DFT) calculations. The simulations reveal a solvent exchange mechanism, where the diffusion of water molecules replaces NMP, driving the formation of the CAB matrix. Bz exhibited faster diffusion and a more uniform distribution compared to Mx, which aggregated into clusters due to its larger molecular size. The analysis of the root mean square deviation (RMSD) and radius of gyration confirmed the faster diffusion of Bz, which adopted a more extended conformation, while Mx remained compact. The phase transformation was driven by the disruption of CAB-NMP hydrogen bonds, while CAB–water interactions remained limited, suggesting that CAB does not dissolve in water, facilitating matrix formation. The molecular configuration revealed that drug–CAB interactions were primarily governed by hydrophobic forces and van der Waals interactions rather than hydrogen bonding, controlling the release mechanism of both compounds. DFT calculations and electrostatic potential (ESP) maps illustrated that the acetyl group of CAB played a key role in drug–polymer interactions and that differences in CAB substitution degrees influenced the stability of drug-CAB complexes. Formation energy calculations indicated that Mx-CAB complexes were more stable than Bz-CAB complexes, resulting in a more prolonged release of Mx compared to Bz. Overall, this study provides valuable insights into the molecular behavior of CAB-based Mx-, Bz-ISG formulations. Full article

Pepper seeds, a key byproduct of pepper processing, are rich in high-quality plant proteins. This study investigated the structural and functional properties of pepper seed protein isolate (PSPI) and optimized the ultrasonic homogenization process for PSPI-based nanoemulsions using response surface methodology (RSM), followed by stability evaluation. The results showed that glutamic acid is the dominant amino acid in PSPI, with a molecular weight range of 10–55 kDa. Some protein subunits were interconnected via disulfide bonds. Functionally, PSPI had lower water-/oil-holding capacities but superior emulsifying activity compared to soy protein isolate (SPI). RSM optimization determined the optimal nanoemulsion parameters within experimental constraints: PSPI 0.53%, ultrasonic power 500 W, and ultrasonic time 130 s, yielding a nanoemulsion with a droplet size of 319 ± 2 nm, consistent with the theoretical prediction (318 nm). The nanoemulsion demonstrated stability under neutral-to-alkaline conditions (pH 7.0–9.0), high ionic strength (Na⁺ concentration ≤ 100 mM), and elevated temperatures (40–100 °C), without phase separation or aggregation. This work supports pepper seed protein utilization and provides insights for plant protein nanoemulsion production. Full article

This article presents an innovative approach as a potential alternative for the reuse of discarded green mussel shells from the fishing and food sectors. This technique entails the use of harmless chemicals and the consumption of energy in an efficient manner to generate shell powder of different dimensions. The shell powder was categorized into three distinct sizes to investigate changes after heat treatment. SEM-EDS was used to analyze particle sizes before calcination and examine the microstructure of heated shell powder. FTIR spectroscopy was conducted to assess the purity of all sizes before and after calcination, showing excellent cleanliness suitable for practical applications. XRD spectroscopy was used to examine the crystal structure, while thermal characteristics and surface color changes during heat treatment were also analyzed due to their impact on final product quality. The variety in particle size enhances the potential for diverse industrial applications. Each size may be suitable for different artificial sand uses, as noted in the conclusion. The proposed method provides both environmental and economic advantages by converting shell waste into a sustainable substitute for artificial sand. It utilizes low-cost, readily available materials and aligns with circular economy principles by reducing shell waste accumulation and dependence on natural aggregates. Full article

As a core task in remote sensing image processing, change detection plays a vital role in dynamic surface monitoring for environmental management, urban planning, and agricultural supervision. However, existing methods often suffer from missed detection of small targets and pseudo-change interference, stemming from insufficient modeling of multi-scale feature coupling and spatio-temporal differences due to factors such as background complexity and appearance variations. To this end, we propose a Differential-Perception-Enhanced Multi-Scale Attention Network for Remote Sensing Image Change Detection (MDNet), an optimized framework integrating multi-scale feature extraction, cross-scale aggregation, difference enhancement, and context modeling. Through the parallel collaborative mechanism of the designed Multi-Scale Feature Extraction Module (EMF) and Cross-Scale Adjacent Semantic Information Aggregation Module (CASAM), multi-scale semantic learning is strengthened, enabling fine-grained modeling of change targets of different sizes and improving small-target-detection capability. Meanwhile, the Differential-Perception-Enhanced Module (DPEM) and Transformer structure are introduced for global–local coupled modeling of spatio-temporal differences. They enhance spectral–structural differences to form discriminative features, use self-attention to capture long-range dependencies, and construct multi-level features from local differences to global associations, significantly suppressing pseudo-change interference. Experimental results show that, on three public datasets (LEVIR-CD, WHU-CD, and CLCD), the proposed model exhibits superior detection performance and robustness in terms of quantitative metrics and qualitative analysis compared with existing advanced methods. Full article

As an intensive eco-agricultural model, the rice–crayfish co-culture (RCC) system has been widely adopted in recent years due to its remarkable advantages in resource use, efficiency, and economic benefits. However, the long-term mechanisms by which this system affects the quantity and stability of soil aggregate, as well as the vertical distribution of soil organic carbon (SOC) within aggregate across soil profiles, remain unclear. This study investigated the effects of varying duration (4 and 8 years) of RCC in Qianjiang City, Hubei Province. Soil samples were collected from six depth layers (0–10 cm, 10–20 cm, 20–30 cm, 30–40 cm, 40–80 cm, and 80–120 cm) to analyze the distribution characteristics of soil aggregate and SOC. The results demonstrated that, compared to the field which used RCC for a duration of 4 years, the field which used RCC for a duration of 8 years significantly reduced bulk density (BD) by 16.3% in the 40–80 cm layer. However, prolonged flooding has led to a 9.6% increase in the BD of the plow pan layer (10–20 cm) due to hydrostatic pressure and mechanical disturbances. Furthermore, the use of RCC for a duration of 8 years significantly enhanced the mass fractions of water-stable aggregates > 2 mm in the 0–80 cm soil layer at 0–10 cm (25.9%), 10–20 cm (30.2%), 20–30 cm (141.8%), 30–40 cm (172.4%), and 40–80 cm (112.9%), and improved aggregate stability throughout the entire soil profile. In terms of SOC distribution, the SOC concentration increased significantly with prolonged RCC usage across all soil layers, particularly in the 0–20 cm layer. The SOC was primarily derived from >2 mm (Large aggregate). Notably, although < 0.053 mm (Silt and clay) constituted a small proportion of the 0–20 cm layer, their SOC concentration reached 15.3–20.55 g kg⁻¹. Overall, extended RCC duration reduced BD in nearly all soil layers, promoted the formation of macro-aggregate, enhanced aggregate stability, and increased the SOC concentration within macro-aggregate, while strengthening the SOC stocks capacity of the 80–120 cm soil layer from 2.58 kg C m⁻² to 4.35 kg C m⁻², an increase of 68.6%. Full article

Starch particles (SPs) were extracted from underutilized wild yam (Dioscorea remotiflora) tubers using two methods: (1) acid hydrolysis (AH) alone and (2) acid hydrolysis assisted by ultrasound (AH-US). The SPs were chemically modified through esterification (using acetic anhydride [AA] and lauroyl chloride [LC]) and crosslinking (with citric acid [CA] and sodium hexametaphosphate [SHMP]). They were subsequently characterized by their yield, amylose content, and structural and physical properties. The yield of particles was 17.5–19.7%, and the residual amylose content was 2.8–3.2%. Particle sizes ranged from 0.46 to 0.55 µm, which exhibited mono-modal and bi-modal distributions for AH and AH-US treatments, respectively. Following chemical modification, yield notably increased, especially with substitution by LC (33.6–36.5%) and CA (32.6–38.7%). Modified SPs exhibited bi-modal particle distributions with micro- and nanoparticles and variable peak intensities depending on the chemical compound used. Unmodified SPs displayed irregular morphologies, showing disruptions (AH) or aggregation (AH-US). Chemical substitutions altered morphologies, leading to amorphous surfaces (CA: AH), clustering (LC), or fragmentation into smaller particles (SHMP) under AH-US treatment. FT-IR analysis indicated a decrease in hydroxyl groups’ peak area (A(-OH)), confirming the substitution of these groups in the starch structure. Crosslinking with CA resulted in the highest degree of substitution (AH: 0.43; AH-US: 0.44) and melting enthalpy (ΔH_f: 343.0 J/g for AH-US), revealing stronger interactions between SPs from both methods. These findings demonstrate that the extraction treatment of D. remotiflora SPs and the type of chemical modifier significantly influence the properties of SPs, underscoring their potential applications as natural biocarriers. Full article