Search Results (12,104)

Generative artificial intelligence and foundation models have changed machine learning by allowing systems to produce readable text, realistic images, and other multimodal content with little direct input from a user. Foundation models are large neural networks trained on very large and varied datasets, and they form the core of many current generative AI (GenAI) systems. Their rapid development has led to major advances in areas like natural language processing, computer vision, multimodal learning, and robotics. Examples include GPT, LLaMA, and diffusion-based architectures, such as models often used for image generation. Systems such as Stable Diffusion show this shift by illustrating how AI can interpret information, draw basic inferences, and produce new outputs using more than one type of data. This review surveys common foundation model architectures and examines what they can do in generative tasks. It reviews Transformer, diffusion, and multimodal architectures, focusing on methods that support scaling and transfer across domains. The paper also reviews key approaches to pretraining and fine-tuning, including self-supervised learning, instruction tuning, and parameter-efficient adaptation, which support these systems’ ability to generalize across tasks. In addition to the technical details, this review discusses how GenAI is being used for text generation, image synthesis, robotics, and biomedical research. The study also notes continuing challenges, such as the high computing and energy demands of large models, ethical concerns about data bias and misinformation, and worries about privacy, reliability, and responsible use of AI in real settings. This review brings together ideas about model design, training methods, and social implications to point future research toward GenAI systems that are efficient, easy to interpret, and reliable, while supporting scientific progress and ethical responsibility. Full article

To improve the efficiency and output power of the twin-screw expander used in natural gas pressure energy recovery, a hybrid NGO-SDERIME algorithm is proposed for structural optimization, with the structural parameters of the male and female rotors selected as the optimization design variables. First, the enhanced Rime Ice Optimization (RIME) algorithm is adopted to perform hybrid improvement on the Northern Goshawk Optimization (NGO) algorithm; then, the stability and superiority of the proposed hybrid algorithm are verified by using a suite of benchmark test functions; finally, the algorithm is applied to the structural optimization of the twin-screw expander, followed by numerical simulation and experimental verification. The results indicate that, compared with other existing algorithms, the proposed NGO-SDERIME hybrid algorithm shows excellent convergence and strong optimization performance. After optimization using this algorithm, the output power of the screw expander increases by 5.5%, the high-speed leakage area is significantly reduced, the isentropic efficiency improves from 75.1% to 78.1%, and the average mass flow rate increases from 18.42 t/h to 18.72 t/h. Full article

During pandemic emergencies, demand for relief supplies in affected areas surges abruptly and evolves randomly and dynamically, resulting in highly asymmetric supply and demand. Ensuring timely and reliable supply requires robust decision-making under risk. This study addresses a stochastic multi-objective location-routing problem (LRP) that simultaneously considers demand uncertainty and travel time variability. A multi-scenario stochastic programming model is developed with three objectives: minimizing total system cost, minimizing total waiting time, and minimizing the composite conditional value at risk (CVaR–Rcomp) to capture tail risks under extreme scenarios. A novel regret-based risk mechanism is introduced to unify temporal and cost dimensions, enabling joint evaluation of uncertainties within a single framework. To solve this challenging high-dimensional problem, a reinforcement learning-enhanced NSGA-III (RL-NSGAIII) is proposed. Specifically, Q-learning generates high-quality initial solutions, which accelerate convergence and improve population diversity for NSGA-III. Case studies demonstrate that the proposed method outperforms traditional evolutionary algorithms in convergence efficiency and Pareto solution quality, while effectively revealing potential risk blind spots. The results provide quantitative decision support and robust optimization insights for emergency logistics networks operating under uncertain conditions. Full article

Public bus systems are vital to achieving sustainable urban mobility in developing countries; yet, the quality of bus stops, a critical interface between users and transit services, remains widely overlooked. This study evaluates bus stop quality in Sulaymaniyah, Iraq, from bus users’ perspectives by integrating importance–performance analysis (IPA) and the customer satisfaction index (CSI) with level of conformity analysis (CR) using extensive, real-world survey data. The objective was to identify priority areas to help improve the quality of public bus stop provision in the city and ensure the most efficient allocation of resources by focusing on the quality attributes that matter most to bus users. The results highlight six critical service quality attributes that require immediate improvement due to their high importance to users and low service quality performance: (i) safety barriers to prevent traffic accidents while waiting at bus stops; (ii) accessibility of bus stops for elderly and disabled users; (iii) availability of signage and timetables/maps; (iv) overall bus stop quality; (v) narrow bus stop platforms; and (vi) waiting time at bus stops. Addressing these gaps is essential to enhance user satisfaction and ensure that users have a safer, more inclusive, and reliable PT experience. This study offers evidence-based recommendations to enhance bus stop design and service quality, thus contributing to improved user satisfaction and increased ridership. More broadly, the results can be applied to other rapidly urbanizing developing cities seeking to provide equitable, safe, and user-centered bus transit systems. Full article

Background: Extracellular vesicles (EVs) are increasingly explored as nanocarriers in drug delivery; however, selecting an appropriate loading strategy for a given small-molecule cargo still relies largely on empirical, resource-intensive parallel screening within EV formulation workflows. Despite the widespread application of passive incubation, electroporation, saponin-mediated permeabilization, freeze–thaw cycling, and sonication, there is currently no mechanistically grounded, descriptor-informed framework that enables rational prioritization of loading methods during the early design stage of EV-based dosage forms, leading to inefficient trial-and-error experimentation. Methods: We assembled a chemically diverse dataset of 21 compounds with experimentally determined loading efficiencies across five EV loading methods and calculated seven mechanistically motivated physicochemical descriptors (LogP, molecular weight, aqueous solubility, hydrogen bond donors/acceptors, polar surface area, and formal charge) for each drug. Separate Elastic Net regression models were trained for each loading strategy. Model performance was evaluated using leave-one-out cross-validation, a predefined external validation set (n = 4), and 50 repeated random train–test splits. The analysis emphasized decision-level ranking of loading methods rather than the precise prediction of absolute efficiencies. The applicability domain was assessed via leverage analysis to define the supported chemical space for prospective implementation in EV-based formulation development. Results: As anticipated for biologically heterogeneous EV systems, continuous regression performance remained modest (LOOCV R² = 0.06–0.41). In contrast, decision-level accuracy for identifying the experimentally optimal loading method was consistently high across validation schemes (internal: 76.5%; predefined external: 75%; repeated random validation: 80.5 ± 16.8%). Mechanical disruption methods (freeze–thaw and sonication) demonstrated comparatively greater predictive stability, while misclassification patterns suggested potential nonlinear behavior for highly polar, ionizable cargos. All compounds resided within the leverage-defined applicability domain, confirming adequate descriptor-space representation. Conclusions: This study establishes a mechanistically interpretable, descriptor-based decision-support framework capable of reliably prioritizing EV loading strategies for small-molecule cargos beyond empirical chance without altering standard protocols. By reframing the modeling objective from high-precision efficiency prediction to robust ranking of candidate methods, the approach offers a practical tool to triage between commonly used techniques, thereby reducing experimental burden in early-stage EV formulation development. The framework provides a quantitative basis for integrating molecular-descriptor-guided method selection into rational EV-based drug delivery design and can be expanded with membrane-specific descriptors and larger datasets. Full article

The accurate measurement and optimization of spatial vitality inside commercial complexes has become crucial for sophisticated urban governance as urban growth moves from rapid expansion to quality-oriented stock augmentation. This research creates a multifaceted assessment methodology that incorporates systemic connectedness (transportation synergy), spatial performance (public activity and social efficacy), and spatial supply (human–land linkages and arrangement). We used a stratified purposive sample of 20 business complexes spread across eight districts in Guangzhou, a typical high-density megacity. In order to understand the underlying mechanisms of spatial vitality, we measured important indicators including the Polycentricity Index (α) and the Spatial Performance Index (β) using a mixed-methods approach that included K-means clustering, multinomial logit regression, and Structural Equation Modeling (SEM). Four important insights are shown by our findings. 1. The paradox of density and efficiency: The notion that high-density development inevitably ensures lively public space is called into question by the lack of a significant linear correlation between the Floor Area Ratio (FAR) and spatial performance (r = 0.32, p > 0.05), despite a core–periphery gradient in development intensity. 2. Structural Supply Demand Mismatch: Although overall spatial performance is strong (β = 0.81 ± 0.07), there is a notable shortfall in cultural and artistic venues, where young adults’ demand (0.27) is 145% greater than supply (0.11). 3. Polycentric Networking vs. Transport Polarization: While spatial structures show a networked polycentric pattern (mean α = 6.40), transportation synergy is affected by core–periphery polarization, which results in “vitality islands” in the periphery. 4. Dual-Path Driving Mechanisms: According to SEM results, cultural spaces have a considerable indirect impact (39.7% mediation) by boosting brand uniqueness and “cultural capital,” while composite plaza spaces have a strong direct effect on commercial performance (γ = 0.682). Based on these findings, we suggest distinct optimization strategies: aging projects need climate-responsive design interventions; growing areas should create family-oriented consumption ecosystems; and core districts should give priority to cultural “IP” integration. For the planning and revitalization of commercial land use in high-density global environments, this study offers a solid analytical framework and practical insights. Full article

This study evaluated the electrochemical and oxidative performance of titanium-supported RuO₂–SnO₂–Sb₂O₅ mixed metal oxide electrodes (hereafter denoted as RuO₂–SnO₂–Sb₂O₅/Ti) for degrading three aniline-based dyes and their mixture using electro-oxidation (EOx), electro-Fenton (EF), and photoelectron-Fenton (PEF) processes. Electrochemical characterization showed quasi-reversible redox behavior and fast electron-transfer kinetics, while SEM, AFM, and EDS analyses revealed a rough surface with fissures and agglomerates that increased the real electroactive area to 4.85 cm², supporting the high catalytic activity. Spectroscopic analyses confirmed the functional groups typical of azo dyes, and RNO assays verified sustained hydroxyl-radical production during electrolysis. Current density was the main operational factor: at 50 mA cm⁻², decolorization exceeded 90% due to enhanced ^•OH generation, whereas higher initial dye concentrations decreased reaction rates because of surface saturation and diffusion limitations. Among the oxidation processes, EF was most effective for Brown KK and Brown 5VR, EOx performed best for Brown NT, and PEF showed a slight advantage for the dye mixture owing to UV-assisted regeneration of reactive species. COD removal followed similar trends, with Brown KK mineralizing fastest and Brown 5VR showing the highest recalcitrance. Analysis of H₂O₂ and active chlorine indicated that EOx favors the accumulation of chlorine-derived oxidants, whereas PEF maximizes H₂O₂ conversion to ^•OH and reduces chlorinated by-products, positioning PEF as the most efficient and environmentally favorable option for treating chloride-containing wastewater. Full article

Industrial digitalization reshapes production processes and can potentially improve carbon productivity by optimizing factor allocation and energy efficiency. Using panel data for 30 Chinese provinces from 2012 to 2022, this study constructs a comprehensive industrial digitalization index with four dimensions and 13 indicators using the entropy method and examines its impact on carbon productivity (GDP per unit of CO₂ emissions). We employ the Dagum Gini coefficient and kernel density estimation to describe regional disparities and their evolution, a dynamic panel threshold model to test the nonlinear role of industrial transformation and upgrading, and a spatial Durbin model to identify spatial spillover effects. The results indicate that industrial digitalization has risen nationwide but remains uneven; industrial digitalization significantly enhances carbon productivity, with stronger effects in the eastern and western regions and in plain areas; the effect exhibits a double-threshold pattern with respect to industrial transformation and upgrading, implying a U-shaped relationship; and industrial digitalization generates positive spatial spillovers. These findings suggest that policy should coordinate digital infrastructure investment with industrial upgrading and regional collaboration to accelerate low-carbon, high-efficiency growth. Full article

Drought is an increasing threat to livelihood security and sustainable development in the Vietnamese Mekong Delta (VMD), particularly in Kien Giang Province. This study examines the spatiotemporal dynamics of meteorological drought from 1992 to 2024 using daily rainfall data from 10 rain gauges. The Standardized Precipitation Index (SPI) was calculated at 3-, 6-, and 12-month timescales to assess short-, medium-, and longer-term precipitation deficits across the province. The results show that the most severe drought events were concentrated in the most recent decade, especially during the 2015–2016 and 2019–2020 dry seasons. Spatial analysis identified clear drought hotspots: the northern coastal zone, including Ha Tien and Hon Dat, exhibited the strongest long-timescale drought signal, while central inland areas such as Go Quao experienced more frequent short-timescale drought conditions. A significant negative relationship was also observed between SPI and the Oceanic Niño Index (ONI), indicating that El Niño conditions intensified drought severity, particularly in coastal areas. These findings highlight the need for spatially differentiated drought adaptation in Kien Giang Province, with stronger emphasis on water storage and water-use efficiency in inland districts and on early warning and integrated drought–salinity management in high-risk coastal zones. Full article

Industrial hemp (Cannabis sativa L.) is versatile and rapidly developing, offering new prospects to producers as a multipurpose crop, yet limited water availability in the Mediterranean area due to climate change makes its sustainable management challenging. Although the plant’s water requirements have been studied, a significant gap remains regarding irrigation management based on the hydraulic properties that govern water movement. The present study elucidates the role of soil hydraulic parameters in water dynamics within the rhizosphere of industrial hemp (Cannabis sativa L., ‘Felina 32’). For this purpose, a pot experiment of three irrigation treatments (100% FC, 80% FC, 60% FC; FC is the field capacity) was set up using two different soil types (clay loam CL and sandy clay loam SCL). SCL soil showed a higher C_max of about 4 cm⁻¹ compared to the C_max of 0.11 cm⁻¹ of CL soil, but dropped drastically within a narrow frame of soil moisture. CL soil resulted in about 12-fold higher diffusivity (D_max ≈ 0.23 cm² min⁻¹) within a wider range of soil moisture compared to the SCL soil (D_max ≈ 0.02 cm² min⁻¹), which facilitated water redistribution at CL, allowing the plant to maximize its water uptake, even at the lowest water input. As a result, the CL soil allowed more flexible scheduling and in contrast, SCL soil necessitated a high frequency irrigation protocol. The integration of hydraulic properties into irrigation planning revealed the potential of CL to apply water to plants efficiently across full and deficit irrigation, showing the peak performance of the irrigation water use efficiency (IWUE) (0.929 g/mm) under the 60% FC regime. The findings provide a framework for linking soil physics–agricultural hydraulics with irrigation strategies in controlled environments. Full article

Precise orbital floor reconstruction requires personalised surgical templates that combine high geometric fidelity with manufacturing efficiency. This study presents and validates the TARMM procedure, developed to optimise the production of polylactide (PLA) templates. A key innovation is the integration of advanced machine learning algorithms (Random Forest) and Mitchell–Netravali interpolation to reduce medical reconstruction artefacts, as well as the implementation of Material Extrusion (MEX) technology with Variable Layer Height (VLH). This strategy minimises the stair-step effect on complex anatomical curvatures while maintaining high process throughput. The results demonstrate that the TARMM procedure ensures a geometric error within ±0.1 mm. A strong linear correlation (r = 0.99) was found between layer height and surface roughness (Sa), indicating that a 0.07 mm layer in critical areas significantly improves template morphology and facilitates the contouring of titanium meshes. The clinical validation across 21 cases confirmed a 30 min reduction in surgical preparation time. The developed method serves as a low-cost, high-precision alternative to photopolymerization technologies, contributing to modern 3D printing applications in maxillofacial surgery. Full article

Human Activity Recognition (HAR) is a pivotal research area for applications such as automated surveillance, smart homes, security, healthcare, and human behavior analysis. Traditional machine-learning approaches often rely on manual feature engineering, which can limit generalization. Although deep learning has improved HAR through automatic representation learning, achieving high detection performance under computational constraints remains challenging. This paper proposes an efficient HAR framework that combines deep learning with hybrid optimization. Surveillance videos are first decomposed into frames, and a keyframe selection stage identifies distinctive frames to reduce redundancy and computational cost while preserving informative content. Motion and appearance features are then extracted using Histogram of Oriented Optical Flow (HOOF) and a ResNet-101 model, respectively, and concatenated into a unified feature representation. Classification is performed using an Inception-based Long Short-Term Memory (Incept-LSTM) network, which is fine-tuned via the proposed Tricky Predator Optimization (TricP) over a restricted, low-dimensional parameter vector. TricP is inspired by predator poaching behavior and the social dynamics of Latrans to enhance exploration and exploitation during search. Experiments on the UCF-Crime dataset show that the proposed method achieves 96.84% specificity, 92.16% sensitivity, and 93.62% accuracy. Full article

Modern high-pressure turbine (HPT) nozzle guide vanes (NGVs) operate under non-uniform inlet conditions, including hot streaks and swirl, which can induce complex flow phenomena and uneven thermal loading. These effects, particularly at the hub-vane corner, can compromise NGV durability, yet the combined influence of swirl and film cooling remains underexplored. The objective of this study is to investigate the aerothermal behaviour of contoured first-stage NGVs under varying swirl intensities and directions to improve understanding of hub and corner thermal protection and failure mechanisms. Steady, compressible RANS simulations were conducted with the k-ω SST turbulence model. A vane with a contoured hub and multiple film cooling rows was designed and analysed under axial and swirling inflows, both clockwise and counter-clockwise, with swirl numbers of Sn = ±0.2 and ±0.4. Axial flow achieved the highest area-averaged film cooling effectiveness (FCE) of 0.617. Negative swirl (Sn = −0.4) improved suction-side corner FCE to 0.215 but reduced pressure-side cooling, whereas positive swirl (Sn = 0.4) improved pressure-side cooling but reduced suction-side FCE to 0.043. Corner temperatures under positive swirl reached 1780 K, consistent with promoting failure, while negative swirl reduced corner temperatures to 1516 K. Aerodynamic losses increased with swirl, with negative swirl generating 5.78% higher total pressure losses than the axial baseline. Swirl altered the corner vortex topology, affecting boundary layer interactions and local heat transfer. These results highlight a trade-off between thermal protection and aerodynamic efficiency, emphasising that optimising NGV performance requires careful design of hub cooling and consideration of swirl direction and intensity. Full article

Longwall paste backfilling mining is a core sustainable green mining technology for coal resources under buildings, railways and water bodies (BRW), yet its large-scale application is severely restricted by the sequential mining–isolation–backfilling–curing operation mode that causes low production efficiency and poor economic feasibility, which hinders the sustainable exploitation of BRW coal reserves and the ecological protection of mining areas. Taking the E1302-B paste backfilling face of Gaohe Coal Mine as the engineering background, this study systematically identified the key efficiency-restricting factors considering the face’s complex geological conditions (maximum roof–floor undulation 300 mm, 72.6% of roof–floor dip angle >1° and irregular cross-section), including low isolation efficiency, cumbersome backfilling process, prolonged paste curing time and insufficient system operation controllability. Technological innovations were carried out from four core dimensions: high-efficiency isolation, high-efficiency backfilling, accelerated curing and intelligent safety control, and a high-efficiency integrated technology system for longwall paste backfilling mining was thus formed, which realizes the synergistic improvement of mining efficiency, economic benefits and sustainability performance. Industrial test validation demonstrated that the technical system significantly boosts the efficiency of isolation, backfilling and solidification in the backfill mining cycle, cutting the time of a single backfill mining operation cycle by 57%. The annual production capacity of the E1302-B face was increased to 0.81 Mt, with a comprehensive backfilling mining cost of 466.63 CNY/t, an annual economic benefit of 108.03 million CNY and a static investment return rate of 48.96%. The E1306 face achieved an even higher annual production capacity of 1.12 Mt with a static investment return rate of 74.94%. This technology system effectively breaks the efficiency and economic bottlenecks of traditional longwall paste backfilling mining, realizes the dual improvement of backfilling mining efficiency and economic benefits, and further releases the ecological, resource and economic sustainability value of paste backfilling mining. It provides technical support and practical approaches for the large-scale application of longwall paste backfilling mining, and lays a solid foundation for the sustainable development of the coal industry under the dual-carbon goal, especially for the balanced development of coal resource exploitation and mining area ecological protection. Full article

Hydroxyapatite (HAp) is a key bioceramic for biomedical applications, but conventional pressureless sintering (PS) requires high temperatures that can promote phase degradation. Here, we compare PS (1100 °C/180 min) and cold sintering process (CSP) (150 °C/450 MPa/30 min) for pure HAp and an HAp composite containing 4 wt.% niobium–phosphate bioglass (BG), using a 2 M H₃PO₄ transient liquid (10 wt.%). CSP increased relative density from 73.10% to 79.92% for HAp and from 68.43% to 83.54% for HAp/BG, representing up to a 22.1% gain compared with PS. One-way ANOVA confirmed a significant effect of processing route/composition on relative density (F(3,24) = 919.69, p < 0.05), and Tukey HSD indicated that all groups differed statistically. SEM revealed a markedly more consolidated and homogeneous microstructure for CSP, particularly for HAp/BG, consistent with enhanced dissolution–reprecipitation and pore filling. XRD showed that PS at 1100 °C led to partial HAp degradation with β-TCP formation, whereas CSP preserved the HAp phase with broader peaks, smaller crystallite size, and higher specific surface area. These results demonstrate CSP as an efficient low-temperature alternative for densifying HAp-based bioceramics, with BG addition further improving consolidation. Full article