Browse Articles

Deep learning–driven image captioning: Progress through transformers and large language models

This paper provides a novel deep learning model for captioning of images by using an advanced vision transformer architecture with a powerful LLM. Proposed models show a significant improvement over traditional CNN-RNN hybrids and existing transformer-based approaches by integrating a unique cross-attention mechanism that enables deep alignment between linguistic context and visual features. We show the superiority of our proposed architecture through extensive evaluation on different datasets like MSCOCO, Flickr30K, and NoCaps. The proposed model consistently shows good performance for leading methods such as GIT, BLIP-2, and CoCa across a comprehensive suite of metrics. On the MS COCO dataset, the BLEU-4, METEOR, and CIDEr scores of proposed models are equal to 0.495, 0.390, and 1.32, respectively. In this paper, we have critically analyzed the key challenges of this field, like enhancing caption diversity, ensuring robust multimodal alignment, and mitigating inherent biases. By providing a new performance level, the proposed model provides a source of reference for the next generation of image captioning systems. The results show the efficiency of our fusion strategy and facilitate the development of techniques that use models that can produce more precise, contextually rich, and human-like image depictions. This work supports SDG 9 (Industry, Innovation, and Infrastructure) by advancing multimodal AI systems, and SDG 4 (Quality Education) by enabling intelligent and accessible image understanding technologies.

Controlled Synthesis of a New Class of Heterostructured Metal Oxides (Cerium, Thorium, Uranium)/Calcium Fluoride Core‐Shell Nanocrystals With Atomically Coherent Interfaces

ABSTRACT Heterostructured nanocrystals (NCs) integrating chemically and/or structurally distinct materials are of significant interest due to their potential to exhibit multiple functionalities or unconventional properties. However, the development of such materials remains limited, primarily due to crystallographic incompatibilities and synthesis challenges. Here, we report the synthesis and structural characterization of a new class of metal oxide‐based (Ce, Th, or U) heterostructures with calcium fluoride featuring core‐shell architectures with precise control over both morphology and shell thickness. Powder x‐ray diffraction (PXRD) and high‐resolution scanning/transmission electron microscopy techniques confirm the formation of high‐quality single‐crystalline particles coupling metal oxide (Ce, Th, or U) and calcium fluoride domains through a structurally coherent but chemically complex intermixed oxide‐fluoride interface—an unprecedented observation in cerium and actinide nanochemistry. These novel materials have been designed to anticipate their future doping with therapeutic alpha particle‐emitting radionuclides (α‐emitters) such as 227 Th or 230 U for locoregional targeted alpha therapy (TAT). It is anticipated that the reported heterostructured core‐shell NCs will offer a promising platform for preclinical investigations to determine the full potential and interest of inorganic core‐shell NCs, especially CeO 2  /   CaF 2 , for TAT application.

Fermi level shifting of CVD-grown MoS2 nanosheets by cobalt doping-driven charge transfer and systematic strain

Precise control of the Fermi level in two-dimensional materials is critical for modulating their electronic structure, thereby facilitating their effective implementation in advanced electronic and optoelectronic applications. Here, we report on the role of cobalt doping in modulating the Fermi level of MoS2 synthesized by the chemical vapor deposition process. Raman spectroscopic results confirmed the presence of p-type dopants and tensile strain in the system through characteristic peak shifts. Photoluminescence studies reinforced the Raman findings by showing progressive quenching of excitonic emission and defect state evolution, consistent with Co-induced p-type doping. X-ray photoelectron spectroscopy validated Co substitution in the MoS2 lattice and provided quantitative insights into bonding states and dopant concentration. Furthermore, Kelvin probe force microscopy measurements demonstrated a clear increase in work function with increasing Co concentration, reflecting a Fermi level shift toward the valence band. These findings highlight the dual role of cobalt doping in inducing strain and charge transfer, thereby offering a tunable strategy for engineering the electronic properties of MoS2 for next-generation nanoelectronics and catalytic applications.

Thermo-environ-economic analysis and multi-criteria optimization with grey wolf optimizer for a biomass-fueled power, hydrogen, and desalinated water tri-generation plant

Cardiorespiratory fitness attenuates the association between psychosocial stress and cardiometabolic risk – Results from the SCAPIS population

Aims Low cardiorespiratory fitness (CRF) and high psychosocial stress can contribute to cardiovascular disease. CRF is a strong predictor of cardiovascular outcomes, yet population-based evidence on whether CRF buffers stress-related risk remains limited. This study aims to (1) examine associations between CRF, stress, and cardiometabolic risk; (2) assess whether CRF moderates the relationship between stress and cardiometabolic risk; and (3) explore whether associations between CRF and cardiometabolic risk are stronger among high-stress individuals. Methods We included 4,207 healthy, middle-aged participants from the Swedish CArdioPulmonary bioImage Study (SCAPIS), conducted 2013–2018. CRF was estimated using submaximal cycle testing (ml/min/kg). Perceived psychosocial stress was measured using a single self-reported item dichotomised into “low” and “high”. Ten cardiometabolic outcomes were assessed, including waist circumference, BMI, and blood pressure. Cross-sectional associations were analysed in R, using t-tests and multiple linear regression. Results Individuals reporting high stress had lower CRF (−1.7 ml/min/kg, p < .001) and less favourable cardiometabolic profiles. CRF moderated the relationship between stress and waist circumference, BMI, systolic- and diastolic blood pressure. For each 1 ml/min/kg higher CRF, the stress-related association was reduced by 0.17 cm in waist circumference (p < .001), 0.06 kg/m 2 in BMI (p < .001), 0.18 mmHg in systolic blood pressure (p = .030), and 0.13 mmHg in diastolic blood pressure (p = .020). Associations were 13–25% stronger in the high-stress group. Conclusion Higher CRF attenuated the association between psychosocial stress and cardiometabolic risk. Promoting physical activity to improve CRF could be important during periods of high stress to counteract stress-related cardiometabolic deterioration.

Effect of Al2O3 capping layer on the crystallization behavior of optical Sb2S3 films

Nanoscale Sb2S3 utilizes the reversible crystalline-amorphous phase transition and is widely used in phase-change memory and near-infrared optoelectronic devices. However, a critical bottleneck persists as its high oxidation susceptibility degrades photoelectrochemical performance. Herein, the crystallization mechanism and oxidation protection pathway of Sb2S3 thin films are clarified by interface engineering. Based on scanning transmission electron microscopy, upon annealing at 300 °C, uncapped Sb2S3 precipitates “pearl chain-like” Sb2O3 along Sb2S3 polygonal domains. The oxidized structure is formed by the permeation and diffusion of environmental oxygen. In contrast, the Al2O3/Sb2S3 heterojunction film annealed at the same temperature exhibits only a single crystalline phase of Sb2S3, confirming the heterointerface as an efficient strategy for blocking oxidation while ensuring the structural integrity and performance stability. Furthermore, the Al2O3/Sb2S3 thin film annealed at 350 °C undergoes a structural transition to a biphasic coexistence structure, where elemental Sb is dispersedly embedded atop the Sb2S3 matrix. This phenomenon originates from a thermal diffusion-dominated crystallization mechanism. These findings provide a universal “heterointerface construction” strategy for oxidation protection of nanoscale phase-change thin films, offering critical guidance for optimizing Sb2S3 device structures and extending their service life.

Smart grid inverter control: integrating RNN, model predictive, and adaptive sliding mode controller for optimal harmonic mitigation

Decisions about risk taking: Elaborate dynamics between guests and hosts of peer-to-peer accommodation during COVID-19

This study explores the influence of COVID-19 on peer-to-peer accommodation, from perspectives of both market performance and guests’ mindset. Combination of qualitative and quantitative data, as well as analysis by smaller time unit, enables more subtle insights into participants’ behavior. Based on Event System Theory, authentic market data is examined to demonstrate the trend of demand and supply of Airbnb housing in New York City. Online reviews are investigated, using Structural Topic Model, to reveal how guests’ focuses transformed. Results show that peer-to-peer accommodation was stricken heavily by COVID-19 and subsequent events, where booking volume shrank more severely, and reacted faster to exogenous environment, compared to housing supply. Guests shifted their attention after COVID-19 happened, caring more about hygiene and social distance. The synchronization of fluctuation among multiple variables validates the assumption that exogenous events successively influenced guests’ attitude and market performance. Apropos of events, stronger incidents exhibited larger influence on market performance, and event timing affected exogenous events’ influence upon the industry. This research enriches Event System Theory and adds to the insights into elaborate dynamics among participants of peer-to-peer accommodation industry during COVID-19. The findings provide hospitality practitioners and governments with reference for future risk management.

Magnetically critically enhanced hybrid hBN quantum thermometer

The boron vacancy (VB−) defects in hexagonal boron nitride (hBN) have shown significant potential in various quantum sensing due to their unique properties. Particularly, the VB− defects are particularly promising for nanoscale quantum thermometers owing to their high temperature dependent zero-field splitting parameter D (−0.8 MHz/K). In order to increase the sensitivity, one method is to increase the slope of the temperature dependent D. In this work, we realize two types of the magnetically critically enhanced hybrid hBN quantum thermometer over a wide temperature range. The slope coefficient dD/dT reached −3.51 and −3.61 MHz/K in the corresponding magnetically critically temperature range for the Cu1Ni3 alloy/gadolinium metal and hBN Hybrid system, with corresponding sensitivities reaching 0.3 and 0.43 K/Hz, respectively. Both the values are about more than three times larger than the bare value, which demonstrates the magnetically critically enhanced effect. This enhancement is attributed to the susceptibility of the magnetic material near its critical temperature. Finally, we compare the spin relaxation rate of the (VB−) in bare and Cu1Ni3 hybrid hBN, and the spin relaxation rate revealed a pronounced peak near the Cu1Ni3 alloy Curie temperature (approximately 323 K). The experiments give the basis for constructing a novel hybrid hBN quantum thermometer, combined with its device compatibility, which could be widely used in practical environments.

“Salvia officinalis extract–conjugated magnetite and selenium nanocomposites showed enhanced antibacterial and anti-biofilm activity against multidrug-resistant pathogens”

Abstract The increased prevalence of multidrug-resistant (MDR) pathogens and biofilm-associated infections underscores the urgent need for alternative antimicrobial strategies. In this study, MDR testing revealed lower resistance in Gram-positive bacteria ( Staphylococcus pasteuri , Listeria monocytogenes , and Bacillus cereus ; 69–77%) than in Gram-negative strains ( Proteus mirabilis , Pseudomonas aeruginosa , and Escherichia coli ; 83–92%). Magnetite nanoparticles (Fe 3 O 4 NPs) and selenium nanoparticles (SeNPs) were synthesized via a green chemical reduction using L-ascorbic acid, followed by their conjugation with Salvia officinalis (SaO) aqueous extract to enhance stability and biological activities. TEM, DLS, XRD, and FT-IR analyses confirmed the nanoscale crystalline structure of the synthesized metal nanoparticles (MNPs) and the successful integration of SaO phytoconstituents into the nanocomposites. Phytochemical characterization further revealed variable adhesion and retention of SaO bioactive compounds within the two nanocomposites. The MIC values of SaO-Fe 3 O 4 NPs (0.03–1) and SaO-SeNPs (0.03–0.5 µg/mL) were not only dramatically lower than the SaO extract (125–500 µg/mL) and MNPs (0.5–25 µg/mL), but also comparable to or even exceeded meropenem (0.25–2 µg/mL). Vigorous antibacterial activity validated by TEM, revealing extracellular and intracellular deformations, in accordance with the growth curve inhibition patterns. The nanocomposites also demonstrated potent antibiofilm activity, achieving 32.5–59.5% inhibition with SaO-Fe₃O₄NPs and 30.2–56.7% with SaO-SeNPs, surpassing meropenem, which exhibited maximum inhibition values of 32.5–47.6%.

Correction: Xiphovelopsis, a new South American genus of Microveliinae (Hemiptera, Heteroptera, Gerridae), with the description of a new species

Differentially enhanced parallel rotating neuron reservoir computing for dynamic gas mixture identification

Accurate identification of mixed gases under dynamically varying concentration conditions is a critical challenge in environmental monitoring and industrial safety. However, the performance of existing methods is severely constrained by sensor cross-sensitivity and weak transient responses. To address these limitations, this paper proposes a differentially enhanced parallel rotating neuron reservoir computing model. The proposed model incorporates dual-path multi-scale differential mechanisms at both the input and output stages, thereby enhancing the sensitivity to dynamic concentration variations from the raw signal domain and the high-dimensional state space, respectively. In addition, a parallel rotating neuron reservoir architecture is employed to efficiently process multi-channel signals and to improve feature diversity. Experiments conducted on the UCI dynamic mixed-gas dataset demonstrate that the proposed model achieves an accuracy of 97.8% in a four-class methane/ethylene classification task and attains normalized mean squared errors of 0.0465 and 0.1359 for methane and ethylene, respectively. These results indicate that the proposed approach provides an effective, training-efficient edge intelligence solution for enhancing the detection efficiency of electronic nose systems.

Chemical characterization and gut microbial response unveil modification of polystyrene polymer in the lesser mealworm

Abstract The massive production of plastics, together with inadequate waste management, has intensified the problem of plastic pollution. Recently, the scientific community has increasingly investigated the interactions between plastic polymers and biological systems, with particular interest in plastivorous insects and their gut-associated microbial communities. The lesser mealworm Alphitobius diaperinus has emerged as a promising model for studying biological interactions associated with polystyrene (PS) modification during gut transit. In this study, two complementary approaches were applied to characterize gut mediated responses to PS. First, chemical analyses were employed to provide evidence of polymer modification following gut transit; micro-FTIR analysis confirmed the presence of PS residues in larval frass and showed detectable structural changes in the recovered material, while GC-MS analysis identified α-methylstyrene and cumyl alcohol as compounds associated with PS chemical modifications and not detected in controls. Second, gut microbiota composition was assessed through metabarcoding analysis using full-length PacBio HiFi sequencing of the 16S rDNA gene. Differences were observed between the microbiota of PS fed larvae and that of the control group under different experimental conditions. Together, these results indicate that PS ingestion leads to measurable chemical modifications of the polymer and shifts in gut microbial composition, providing insight into insect–microbe interactions associated with PS transformation.

Long-term farming and cropping systems with contrasting nitrogen forms and input diversity influence soil prokaryotic diversity in the central highlands of Kenya

Background Understanding how farming systems management influences soil microbial communities is essential for advancing sustainable agriculture in tropical regions. Long-term experiments provide valuable opportunities to assess how cumulative management practices shape soil microbial diversity and community composition. Methods We investigated prokaryotic communities after 15 years of continuous management in the Long-Term Farming Systems Comparison Trial (SysCom-Kenya) at two contrasting sites (Chuka and Thika) in the Central Highlands of Kenya. Four systems were evaluated: conventional low-input (Conv-Low), conventional high-input (Conv-High), organic low-input (Org-Low), and organic high-input (Org-High). Soil samples were collected at key crop growth stages (vegetative, reproductive, and maturity) of maize, baby corn, and potato. Prokaryotic diversity and community composition were characterized using 16S rRNA gene amplicon sequencing, and soil chemical properties were analyzed to explore potential abiotic drivers. Results Prokaryotic community composition and diversity varied primarily with site and farming system, with secondary variation across crop growth stages. Across all systems, communities were dominated by members of the phyla Proteobacteria and Actinobacteria, followed by Acidobacteria, Firmicutes, and Chloroflexi. Organic systems, particularly organic high-input, tended to support higher richness and evenness than conventional systems, while low-input systems consistently exhibited lower prokaryotic richness and diversity than high-input systems. Diversity generally increased toward later crop growth stages, although phenological effects were variable. Canonical correspondence analysis identified soil pH, ammonium-N, and available phosphorus as important correlates of community structure, especially at the drier Thika site. Taxon-specific enrichment patterns differed across systems and crop stages, indicating compositional differentiation rather than functional dominance. Conclusion Our findings indicate that long-term management intensity and organic input diversity exert a stronger influence on soil prokaryotic communities than short-term crop phenology. Despite limitations from sample pooling, this study provides novel evidence from sub-Saharan Africa that diversified organic input management can enhance soil microbial diversity and potential resilience, supporting sustainable soil management in tropical farming systems.

Synergistic bandgap and heterojunction engineering in YbSnO thin films for high-performance self-powered solar-blind photodetection

Exploring semiconductor materials with suitable bandgaps and chemical stability is crucial for constructing solar-blind ultraviolet (SBUV) photodetectors with high stability, high responsivity, and high detectivity. However, the scarcity of suitable material systems has significantly hindered further breakthroughs in device performance. Rare-earth oxide Yb2O3, with its wide intrinsic bandgap (∼4.9 eV) and high dielectric constant, is theoretically promising for SBUV detection. Nevertheless, its practical application has been limited by an excessively large bandgap and poor electrical conductivity. Herein, we propose a synergistic strategy combining bandgap engineering and a double heterojunction design to fabricate a p-Gr/i-YbSnO/n-SiC photovoltaic-type SBUV photodetector. The active YbSnO film was realized by co-sputtering SnO2 into Yb2O3, which effectively narrows the bandgap to 4.42 eV and redshifts the absorption onset to 280 nm. By integrating monolayer graphene (Gr) as a transparent top electrode, a p-Gr/i-YbSnO/n-SiC double heterojunction was constructed. Leveraging the dual built-in electric fields, the device achieves remarkable self-powered performance under 255 nm illumination: an open-circuit voltage of 0.33 V, a responsivity of 18.41 mA/W, an external quantum efficiency of 8.96%, and a high specific detectivity of 2.31 × 1012 Jones, along with a rejection ratio (R255 nm/R340 nm) of 282. This work achieves precise bandgap control of rare earth oxide, significantly optimizes the device performance of Yb2O3-based semiconductors for SBUV detection, and provides key technical support for the practical application of such materials in miniaturized and integrated optoelectronic chips.

A Caputo fractional-order SEIHRD model for Ebola: theoretical analysis, sensitivity, bifurcation, and numerical simulations

Abstract This study develops and analyzes a Caputo fractional-order SEIHRD model to investigate the transmission dynamics and control of Ebola virus disease. The model ensures positivity, boundedness, and invariance of feasible regions for all solutions. Rigorous analysis establishes local and global existence, uniqueness, and well-posedness of the system. The basic reproduction number $$R_0$$ is derived, with stability analysis of disease-free and endemic equilibria revealing a forward transcritical bifurcation at $$R_0 = 1$$ . Sensitivity analysis identifies key parameters significantly influencing $$R_0$$ and endemic infection levels. Numerical simulations using both the Fractional Runge-Kutta scheme and the Fractional Differential Transform Method demonstrate the pronounced impact of the fractional order on system stability and persistence, with the Runge–Kutta method providing superior accuracy. These results highlight the critical role of fractional-order modeling in capturing memory effects in epidemic processes and suggest the efficacy of fractional calculus in enhancing epidemic predictions. The results show that fractional-order dynamics capture Ebola’s persistence and memory effects, providing a framework for control strategies. This also points toward incorporating fuzzy fractional approaches to better address parameter uncertainty, offering a robust framework for future extensions in epidemic modeling and control strategies.

Correction: Understanding street protests: From a mathematical model to protest management

Temperature- and frequency-dependent dielectric behavior in epitaxial CaSnO3 films

We systematically examine the dielectric response of epitaxial CaSnO3 films grown by hybrid molecular beam epitaxy. Metal-insulator-metal capacitor structures consisting of Nb:SrTiO3 (001)/t-nm CaSnO3/100-nm Pt (top electrode) were fabricated to characterize the dielectric properties of CaSnO3, including the dielectric constant (κ) and loss tangent across wide temperature (1.8–400 K) and frequency ranges (20 Hz–1 MHz). Films exhibited a dielectric constant of ∼17 and a low loss tangent (<0.02), independent of thickness and electrode geometry. The minimal variation with temperature and frequency highlights CaSnO3's potential as a robust, low-loss, high-κ dielectric for advanced oxide electronics and gate insulator applications.

An energy-efficient Flash-SAR ADC with two-step flash structure in a 0.18 μm CMOS process

Study on the microstructure and properties of laser-cladded AlCoCrFeNiCu0.5-xNbₓ high-entropy alloy coatings

In this study, the high-entropy alloy AlCoCrFeNiCu0.5 − xNbx (x = 0, 0.1, 0.2, 0.3, 0.4, 0.5) is selected as the research subject. Coatings with varying Cu/Nb ratios are fabricated by laser cladding. The phase composition, microstructure, microhardness, tribological behavior, and corrosion resistance are systematically investigated. The results reveal that when x = 0.2, the coating exhibits a dual-phase structure consisting of FCC and Laves phases. A large cellular structure forms and maintains good microstructural continuity. The coating shows a low friction coefficient of 0.327, and no significant spalling is observed on the worn surface. In both 3.5% NaCl and 0.5 M H₂SO₄ solutions, the coating exhibits the most positive corrosion potential, the lowest corrosion current density, and the highest polarization resistance, indicating excellent electrochemical stability.