As an important level in the administrative system, counties have abundant natural and agricultural resources and play a key role in promoting green development strategies. This article uses the entropy method to construct a comprehensive evaluation index system for green and low-carbon development from five dimensions: green innovation, green coordination, green efficiency, green openness, and green sharing. It uses econometric methods such as Moran’s index and kernel density estimation to empirically analyze the level of green and low-carbon development in Chinese counties, differentiation characteristics, and spatial effects. Furthermore, geographically weighted regression model is introduced to explore and analyze the driving factors of various variables on the differences of green development in Chinese counties. The study finds the following: From a temporal perspective, the overall level of green development in Chinese counties shows a fluctuating upward trend; From the perspective of spatial evolution, the green development of Chinese counties presents a regional distribution characteristic of “moderate distribution in west, high in the east and low in west”; The level of green development in counties has a significant spatial positive correlation. The driving factors of economic development level, green technology innovation, and government support mainly have a positive impact on the green development of Chinese counties, while the quality of human resources mainly has a negative impact. The research results emphasize that there is obvious spatial heterogeneity in the impact of each driving factor. By implementing differentiated green development strategies to alleviate resource and environmental problems between counties, deepening the concept of green and low-carbon development, optimizing the industrial layout structure of counties, and adhering to the strategy of innovation driven development, suggestions can help counties achieve sustainable development goals.

This study develops a practical framework for forecasting long-term drought conditions in Karaman Province, a semi-arid region of Turkey, where accurate climate information is vital for water planning and agriculture. Since the area has limited rainfall records and strong year-to-year fluctuations, traditional modeling approaches often fall short. To better capture local conditions, drought intensity was defined using a simple monthly wetness anomaly measure based directly on precipitation; here, positive values indicate wetter months and negative values indicate drier ones. This makes the method suitable for regions where detailed hydrological data are scarce. Rainfall observations from 1965 to 2011 were expanded using a combination of kernel density estimation and Cholesky-based correlation reconstruction. These steps preserved the main statistical and temporal patterns of the original data while increasing sample diversity. The enriched dataset was then used to train artificial neural networks to predict both precipitation and drought intensity. The models reached R2 values of 0.76 and 0.72, with mean absolute errors of 12.8 mm and 28.4%, which represents an improvement of roughly 10–15% over traditional statistical methods. They were also able to capture the seasonal and year-to-year variability that strongly affects drought conditions in the region. To understand what drives the predictions, the model was examined with LIME, which consistently highlighted lagged rainfall and seasonal indicators as the most influential inputs. A walk-forward validation approach was also used to mimic real forecasting conditions and demonstrated that the model remains stable when projecting into the future. Overall, the proposed framework offers a reliable and practical basis for early-warning efforts and drought-management strategies in semi-arid regions like Karaman.

Abstract. Density of seawater is a critical property that controls ocean dynamics. Previous works suggest the use of the δ18O calcite of foraminifera as a potential proxy for paleodensity. However, potential quantitative reconstructions were limited to the tropical and subtropical surface ocean and without an explicit estimate of the uncertainty in calibration model parameters. We developed the use of the δ18Oc of planktonic foraminifera as a surface paleodensity proxy using Bayesian regression models calibrated to annual surface density. Predictive performance of the models improves when we account for inter-species specific differences. We investigate the additional uncertainties that could be introduced by potential evolution of the δ18Oc-density relationship with time – from the last glacial maximum (LGM) to the preindustrial (PI) – through the combination of past isotope enabled climate model simulations and a foraminiferal growth module. We demonstrate that additional uncertainties are weak globally, except for the Nordic Seas region. We applied our Bayesian regression model to LGM and Late Holocene (LH) δ18Oc foraminifera databases to reconstruct annual surface density during these periods. We observe stronger LGM density value changes at low latitudes compared to mid latitudes. These results will be used to evaluate numerical climate models in their ability to simulate ocean surface density during the extreme climatic period of the LGM. The new calibration has great potential to reconstruct the past temporal evolution of ocean surface density over the Quaternary. Under climates outside the Quaternary period and in ocean basins characterized by anti-estuary circulation, like the current Mediterranean Sea and Red Sea, our calibration could provide density estimates with larger uncertainty, a point that requires further investigations.

Amid China’s shift from a model of urban “incremental expansion” to one focused on “stock optimization”, the renewal of streetscapes has taken center stage as a critical approach to improving the human experience within urban environments. However, empirical insight into how visual interventions affect aesthetic perception across different road types remains notably limited. This study addresses that gap through a spatiotemporal investigation of Zhengzhou’s streetscape transformations between 2017 and 2022. Major roads were categorized into four functional types—freeway, under-freeway, regular road, and tunnel—to better capture perceptual variation. Leveraging a Fully Convolutional Network (FCN), we extracted nine visual components from historical street views and paired them with crowd-sourced “beauty” ratings from the MIT Place Pulse 2.0 dataset. Statistical analyses, including paired t-tests and Kernel Density Estimation (KDE), indicated marked improvements in perceived beauty following renewal, with the exception of tunnel segments. Through Random Forest (RF) regression and SHapley Additive exPlanations (SHAP) interpretation, greening emerged as the most influential driver of aesthetic enhancement—most prominently on regular roads (SHAP = 2.246). The impact of renewal was found to be context-specific: green belts were most effective in under-freeway areas (SHAP = +0.8), while improvements to pavement (SHAP = +0.97) and street vitality were key for regular roads. Notably, SHAP analysis revealed non-linear relationships, such as diminishing perceptual returns when green coverage exceeded certain thresholds. These findings inform a “visual renewal–perceptual response” framework, offering data-driven guidance for adaptive, human-centered upgrades in high-density urban settings.

Physiological time series reflect the underlying behavior of physiological systems. In this paper, we introduce a novel patching with sequential updating for Bayesian nonparametric spectral estimation (PBNSE) to enhance spectral estimation and interpretation of imperfect physiological time series with fragmented, noncontiguous segments. PBNSE incorporates four key strategies: (1) modeling patches as patch-specific Gaussian processes (GPs); (2) patch-dependence, where each patch involves a joint GP with a shared kernel, capturing both observation and spectral dependencies across all patches; (3) sequential parameter shift that transfers knowledge between patches while maintaining computational traceability; and (4) aggregating patch-level posterior spectra into a unified power spectral density (PSD) estimate and computing the expectation of the PSD in a closed form. Extensive experiments demonstrate significant improvements in spectral accuracy and robustness compared to state-of-the-art methods such as BNSE, multitaper, periodogram, Lomb-Scargle, functional kernel learning (FKL), and variational sparse spectrum (SVSS). PBNSE addresses key challenges in physiological signal analysis, including irregular sampling, incomplete signal, and varying noise. The widespread adoption of PBNSE in physiological signal research has the potential to enhance the accuracy of spectral estimation and improve the robustness of interpreting complex, real-world physiological time series.

This study develops a generalized Newton method to address the nonlinear Kolmogorov forward equation (KFE) under the local stochastic volatility (LSV) framework. Analytical convergence conditions are derived via the geometric series theorem, and empirical validation is conducted using 15 years of monthly crude oil spot price data (2011–2025), with the parameters set using maximum likelihood estimation. Sensitivity analyses confirm the stable convergence of the iterative scheme under realistic scenarios while also identifying parameter ranges that may lead to divergence. These findings demonstrate that the proposed methodology provides a tractable and accurate approach to probability density estimation in commodity markets, with a clear potential for extension to multi-dimensional settings and richer datasets.

Context Understanding how animals navigate fragmented landscapes is critical for conservation planning, particularly in agricultural regions where native vegetation is limited. The short-beaked echidna (Tachyglossus aculeatus) remains relatively common in these environments, yet little is known about how fragmentation influences its space use and movement. Aims This study aimed to investigate how habitat composition and fragmentation affect echidna movement, home range size, and path structure in a working agricultural landscape. Methods We GPS-tracked 10 adult echidnas (5 males, 5 females) across three farms (~35 km²) in the Liverpool Plains, NSW, between July 2022 and December 2023. Home ranges were estimated using 95% autocorrelated kernel density estimation (AKDE). Habitat selection, movement rates, and path sinuosity were analysed with respect to habitat type (woody vs. open) and fragmentation metrics, including connectivity, proximity, clumpiness, and patch shape complexity, derived using FRAGSTATS. Key Results Despite open habitats being more abundant, echidnas strongly selected woody patches. Movement rates were faster in open areas, likely reflecting travel between resources. Home range sizes varied (0.55–7.76 km²), with connectivity emerging as the strongest predictor. Echidnas in more connected landscapes had significantly smaller home ranges, travelled shorter distances, and moved along straighter paths. Greater proximity variation led to more tortuous paths, suggesting local patch context affects movement complexity. Other fragmentation metrics had limited explanatory power. Conclusions Echidnas exhibited strong preferences for woody habitats and altered their movement behaviour in response to habitat fragmentation. Higher connectivity reduced home range size and movement distances, while local patch configuration influenced path structure. Implications These findings underscore the importance of retaining and restoring woody vegetation and improving functional connectivity in agricultural landscapes. Conservation strategies that prioritise habitat continuity and fine-scale heterogeneity will better support echidna persistence and movement in modified environments.

ABSTRACT This study introduces an innovative approach, the Predictive Approach, employing the Temporal Convolutional Network (TCN) algorithm to estimate traffic density. We used naturalistic vehicle trajectories captured by drones at a three-way signalized intersection in Athens, Greece, as part of the pNEUMA initiative. This method calculates the densities of input approaches at intersections with non-uniform MPRs, using these predictions to estimate the target approach density. With accuracy ranged from 92% to 95%, using the Predictive Approach showed that improving traffic density predictions can be achieved through factors such as accounting for MPR variations over time and between different intersection approaches while considering practical scenarios. Results also highlighted that excluding Signal Phase and Timing (SPaT) data in certain cases can enhance model performance. It offers practical applications in optimizing traffic flow and reducing congestion in smart cities and traffic control centers, particularly when rapid and real-time computations are required.

Unbiased and accurate estimation of bird density are prerequisites to monitor the impact of urbanization on avian communities. Synanthropic birds are reliable indicators of landscape modification in small tropical cities with rural-urban ecological settings. We conducted 183 fixed-width point counts to record avian communities along the rural-urban gradient in Mirzapur and Bhadohi. We applied hierarchical distance-sampling to estimate the summer density of 35 bird species across eight foraging guilds in response to vegetation, land cover, human activity and housing type, accounting for detection probability as a function of weather. Twenty-seven species (77%) showed differences in density across the landscape gradient. The density of frugivores was highest in urban gradient, a carnivore was highest in rural, and a scavenger was highest in the semi-urban gradient thereby supporting the resource concentration hypothesis. Insectivores, granivores and omnivores showed variable density patterns along the gradient. The relatively lower density of synanthropic birds in semi-urban and urban fringes indicates the need for enhanced green space and vegetation structure along intermediate landscape gradient. The habitat associations and population sizes of synanthropic birds are useful for landscape managers and local stakeholders to maximize avifaunal diversity in the intermediate landscapes of the rural-urban continuum in Uttar Pradesh.

ABSTRACTReliable estimates of population density are essential for the conservation of apex predators such as the jaguar (Panthera onca), particularly in peripheral regions of their distribution where existing data are insufficient to guide effective management. In Mexico, northeastern landscapes remain underrepresented in jaguar research, limiting the development of context‐specific conservation strategies. To address this gap, we conducted a camera trap survey in the El Cielo–Sierra de Tamalave biological corridor, a transitional zone located at the northeasternmost limit of the species' range. Over a 91‐day sampling period, we deployed 104 cameras across 52 paired stations and applied a random thinning spatial capture–recapture model (rt‐SCR), which integrates both identified and unidentified photographic detections. This represents the first application of rt‐SCR to jaguar data in Mexico. The model yielded a density estimate of 1.29 (0.93–1.70) individuals per 100 km2, with adequate goodness‐of‐fit across multiple detection metrics. Despite low detection rates, the rt‐SCR framework allowed for robust inference by maximizing data use and mitigates the loss of precision associated with excluding unidentified detections. Our findings provide a baseline for future monitoring in northeastern Mexico and demonstrate the utility of rt‐SCR models in data‐limited contexts. These results support the implementation of localized conservation actions and long‐term monitoring programs in peripheral jaguar habitats, where population viability may depend on maintaining ecological continuity and minimizing anthropogenic pressures.

Frequency-Domain Approach to Automated and Efficient Multivariate Kernel Density Estimation for Probabilistic Modeling

Crack density estimation in rock structures using machine learning techniques

Kernel density estimation with a Markov chain Monte Carlo sample

ABSTRACTHandling missing values is a common challenge in biostatistical analyses, typically addressed by imputation methods. We propose a novel, fast, and easy‐to‐use imputation method called missing value imputation with adversarial random forests (MissARF), based on generative machine learning, that provides both single and multiple imputation. MissARF employs adversarial random forest (ARF) for density estimation and data synthesis. To impute a missing value of an observation, we condition on the non‐missing values and sample from the estimated conditional distribution generated by ARF. Our experiments demonstrate that MissARF performs comparably to state‐of‐the‐art single and multiple imputation methods in terms of imputation quality and fast runtime with no additional costs for multiple imputation.

Learning unnormalized statistical models (e.g., energy-based models) is computationally challenging due to the complexity of handling the partition function. To eschew this complexity, noise-contrastive estimation (NCE) has been proposed by formulating the objective as the logistic loss between the real data and the artificial noise. However, previous research indicates that NCE may perform poorly in many tasks due to its flat loss landscape and slow convergence. In this paper, we study a direct approach for optimizing the negative log-likelihood of unnormalized models through the lens of compositional optimization. To tackle the partition function, a noise distribution is introduced such that the log partition function can be expressed as a compositional function whose inner function can be estimated using stochastic samples. Consequently, the objective can be optimized via stochastic compositional optimization algorithms. Despite being a simple method, we demonstrate it is more favorable than NCE by (1) establishing a fast convergence rate and quantifying its dependence on the noise distribution through the variance of stochastic estimators; (2) developing better results in Gaussian mean estimation by showing our method has a much favorable loss landscape and enjoys faster convergence; (3) demonstrating better performance on various applications, including density estimation, out-of-distribution detection, and real image generation.

ABSTRACT The high acquisition cost of synthetic aperture radar (SAR) imagery has been a persistent obstacle to advanced deep learning researches. Recently, image translation techniques have emerged as promising solutions for augmenting SAR datasets by translating readily available optical images into SAR-like representations. However, the substantial stylistic differences between optical and SAR images pose significant challenges in accurately extracting optical image semantics and replicating SAR image styles, especially when co-registered data is unavailable. To address this challenge, we propose an unpaired optical-to-SAR image translation (O2SIT) method, named extract-and-transform generative adversarial network (ET-GAN). First, we introduce cascaded coordinate attention (CA) bottleneck blocks that enhance the positional information of feature maps, thereby precisely extracting optical image semantics. Second, to better capture SAR style characteristics, we employ histograms as auxiliary supervision by constructing a differentiable histogram using kernel density estimation and global average pooling. On this basis, the squared earth mover distance is adopted as an additional loss to guide the generator in producing synthetic images with pixel distributions similar to real SAR images. Experimental results on SEN12, WHU-SEN-City, and GaoFen aircraft detection (GF-AD) dataset demonstrate that ET-GAN achieves competitive SAR image generation performance compared to other state-of-the-art methods, with PSNR of 17.11 on SEN12 and FID of 168.16 on GF-AD. Transfer learning results demonstrate that the images generated by ET-GAN can bring about 3% accuracy improvement to SAR aircraft detection.

Dengue fever continues to pose a significant public health threat in Malaysia, particularly in peri-urban districts undergoing rapid residential expansion. However, the extent to which housing structures influence dengue transmission remains poorly quantified. This study investigated the temporal, proportional, and spatial dynamics of dengue across five housing categories; landed properties, high-rise residential units, traditional/rural houses, institutional quarters, and others in Kuala Selangor from 2020 to 2024. A total of 5,426 laboratory-confirmed dengue cases obtained from the national e-Notifikasi system were geocoded and classified by housing type. Temporal trends were examined using weekly epidemic curves, proportional contributions were calculated for each housing category, and spatial clustering was assessed using Kernel Density Estimation (KDE) in ArcGIS Pro. Landed properties were the dominant transmission environment, contributing 73.4% of all cases and consistently driving major seasonal peaks during epidemiological weeks 20-35, coinciding with the southwest monsoon. High-rise residential areas accounted for 16.1% of cases and exhibited persistent low-level transmission throughout the year, indicating a potential role as an endemic reservoir between epidemic cycles. Traditional/rural houses (5.5%), institutional quarters (3.4%), and other categories (1.6%) contributed only sporadically. KDE mapping revealed persistent hotspots in central and southern Kuala Selangor, primarily within peri-urban landed housing estates, with smaller recurrent clusters in high-rise complexes. These findings demonstrate that housing typology is a critical determinant of dengue transmission risk. Landed properties amplify monsoon-driven outbreaks through abundant outdoor breeding habitats, while high-rise buildings sustain inter-epidemic transmission via sheltered, indoor breeding sites. Integrating housing-specific intelligence into Malaysia's Integrated Vector Management (IVM) framework can enable more targeted, proactive, and spatially adaptive dengue prevention strategies.

Understanding the spatial distribution and associated contextual factors of historic and cultural districts is essential for heritage conservation and land-based spatial planning. Previous studies have largely focused on site-level authenticity and development models, while national-scale quantitative analyses remain limited, particularly in the Chinese context. This study constructs a nationwide dataset of 1212 historic and cultural districts across 31 provincial-level regions in China and applies a GIS-based framework integrating Global Moran’s I, kernel density estimation, and standard deviation ellipse analysis. The results reveal significant spatial clustering broadly aligned with the Hu Huanyong Line, with high-density concentrations in major urban agglomerations such as the Yangtze River Delta and an overall orientation along an east–west axis at the national scale. Further analysis suggests that these spatial patterns are broadly consistent with natural geographic conditions and long-term historical–cultural contexts; river systems, dynastic capital construction, economic development, and population migration provide important interpretive backgrounds for understanding regional differentiation. By providing a national-scale quantitative characterization of designated historic and cultural districts, this study offers spatial evidence to support more differentiated, land-oriented conservation and spatial planning strategies and to facilitate comparable spatial-heritage analytics in other regions where designation lists exist.

New-build gentrification, a type of gentrification which is connected to newly built development, has radically transformed the appearance of neighborhoods across the United States. However, the literature is lacking discussion on the built component of the new-build gentrification process, which can lead to inaccurate maps and projections of gentrification trends. Recent advancements in machine learning (ML), specifically computer vision models that apply neural network "deep mapping" algorithms, have found application in the research for their ability to track changes in urban streetscapes. In our research, we trained machine learning models to identify new-build development with architectural traits that reflect visual cues of gentrification according to local residents. With Philadelphia as our study area, we drew on the insight of community-based focus groups to identify characteristics that denote new-build gentrification for the city. We compared our audit of new-build gentrification development with municipal permit License and Inspections (L&I) data, using Kernel Density Estimate (KDE) maps to visualize the spatial trends of both datasets. Our final fine-tuned ResNet-50 model achieved an 84.0% test accuracy and an 84.0% Area Under the Curve (AUC) score. Our research contributes a novel mixed-methods approach that integrates community input with Artificial Intelligence (AI) to identify locally-specific gentrification traits.

. This article introduces a robust semiparametric mixture of regression (SMR) model utilizing symmetric α-stable (SαS) distribution to address limitations of traditional normal-error SMR frameworks when handling outliers or heavy-tailed data. The proposed model maintains SMR’s core structure with non parametrically estimated mixture weights via kernel density estimation and cluster-specific parametric regressions. The SαS distribution is a generalization of the normal distribution with an additional parameter called α. The normal distribution is obtained when α = 2, and for α < 2, it exhibits heavier tails and unbounded variance, i.e., the variance is infinite when α < 2, thereby providing extreme outlier resilience compared to finite-variance alternatives (e.g., t-distributions with degrees of freedom greater than 2). Parameter estimation employs an expectation-maximization (EM) algorithm for regression coefficients and scale parameters, while α is estimated using Monte Carlo integration and interpolation. Model performance is evaluated against existing mixture regression methods through comprehensive simulations and real-data analyses.