Data Constraints Research Articles

Great expectations: understanding perceptions of the affordances and constraints of dashboard data

ABSTRACT Drawing upon literature and theorising in relation to the sociology of expectations and critical data studies, we elaborate perceptions of the nature of dashboard data in one state educational jurisdiction in Australia. Our research utilises interviews with senior educational bureaucrats, who engaged with a new dashboard, and the data it generated, during the initial stages of the development and implementation of the dashboard. This included personnel involved in the technical development of the dashboard, through to those who were responsible for engaging with such data to help facilitate enhanced school organisational, teaching and learning practices across the state. The research reveals that at the same time as data provided through the dashboard were understood by educators as more current and potentially beneficial for making sense of students' learning, such data also simultaneously conveyed information that was potentially not as accurate, accessible, or timely as desired. Consequently, such dashboards are complex sites of ‘great expectations', in which positive futures about dashboard data are expressed, even as reservations are simultaneously articulated. While such dashboards enhance access of information to educators in a more ‘timely’ fashion, there is a need for further scrutiny into such platforms and the veracity of the data generated.

Hybrid learning strategies: integrating supervised and reinforcement techniques for railway wheel wear management with limited measurement data

Train wheel wear significantly impacts wheel-rail interaction forces and is an unavoidable issue in the railway industry. This study focuses on regular wear, specifically changes in wheel profiles such as tread wear, flange height, and flange thickness. Effective wheel wear management is crucial for maintaining the reliability, safety, and efficiency of rail systems. However, regular measurement of wheel profiles is often limited by constraints such as dense traffic, budget, time, and remote assets, which reduces the effectiveness of traditional maintenance approaches. This study proposes a hybrid learning strategy combining supervised and reinforcement learning techniques to optimize train wheel wear management under these constraints and achieve predictive maintenance. The supervised learning model, developed from validated simulations, predicts wear progression, while reinforcement learning improves maintenance decision-making using basic operational data without regular measurements. Various machine-learning techniques are explored and fine-tuned to identify the best models for preventing faulty wheels without the need for frequent inspections. By integrating these two learning approaches, the framework enhances the accuracy of wear predictions and optimizes maintenance schedules, reducing the risk of over-maintenance or unexpected failures. This integrated model addresses challenges such as system complexity, limited data, and cost-effectiveness in the industry. In terms of supervised learning, the R2 for tread wear prediction improves from 0.94 to 0.95 compared to previous studies, and the model, when integrated with reinforcement learning, significantly reduces defects based on wear and irregular wheel dimensions. This research is the first to integrate supervised and reinforcement learning specifically for train wheel wear management under limited measurement data constraints, offering a breakthrough compared to traditional methods that rely on regular inspections. The study provides significant benefits for the railway industry, including reduced maintenance costs, improved maintenance efficiency, lower defect rates, reduced possession and inspection time, and enhanced passenger comfort and safety.

Enhancing Activation Energy Predictions under Data Constraints Using Graph Neural Networks.

Accurately predicting activation energies is crucial for understanding chemical reactions and modeling complex reaction systems. However, the high computational cost of quantum chemistry methods often limits the feasibility of large-scale studies, leading to a scarcity of high-quality activation energy data. In this work, we explore and compare three innovative approaches (transfer learning, delta learning, and feature engineering) to enhance the accuracy of activation energy predictions using graph neural networks, specifically focusing on methods that incorporate low-cost, low-level computational data. Using the Chemprop model, we systematically evaluated how these methods leverage data from semiempirical quantum mechanics (SQM) calculations to improve predictions. Delta learning, which adjusts low-level SQM activation energies to align with high-level CCSD(T)-F12a targets, emerged as the most effective method, achieving high accuracy with substantially reduced data requirements. Notably, delta learning trained with just 20-30% of high-level data matched or exceeded the performance of other methods trained with full data sets, making it advantageous in data-scarce scenarios. However, its reliance on transition state searches imposes significant computational demands during model application. Transfer learning, which pretrains models on large data sets of low-level data, provided mixed results, particularly when there was a mismatch in the reaction distributions between the training and target data sets. Feature engineering, which involves adding computed molecular properties as input features, showed modest gains, particularly in thermodynamic properties. Our study highlights the trade-offs between accuracy and computational demand in selecting the best approach for enhancing activation energy predictions. These insights provide valuable guidelines for researchers aiming to apply machine learning in chemical reaction engineering, helping to balance accuracy with resource constraints.

Parking, travel behavior, and working from home

We examine parking policy’s potential to influence people’s travel behavior, including their decision to work from home. Drawing on the California Household Travel Survey, we first show that most households have residential parking included in the rent or purchase price of their home, that most employed individuals have free parking at work, and most vehicle trips end with a free parking space, usually off-street. Furthermore, we confirm that most cars are mostly parked; the median household vehicle in California spends 23 hours a day parked. Using regression models, we uncover strong associations between choosing to drive and having free parking at work or home. We find that households with bundled parking are more likely to drive, and less likely to use transit. We further find that employees with free parking at work are more likely to drive for their commutes. Finally, we estimate regressions that analyze the decision to work from home. Data constraints make these regressions less conclusive, but our results suggest, consistent with previous literature, that working from home is associated with more, rather than less, vehicle travel.

Epidemiological Characteristics and Prognostic Scoring in Toxic Epidermal Necrolysis and Stevens-Johnson Syndrome: Insights from a 17-Year Burn Center Experience.

Background and Objectives: Toxic epidermal necrolysis (TEN) and Stevens-Johnson syndrome (SJS) are rare yet life-threatening dermatologic conditions characterized by severe skin and mucous membrane involvement. Accurate prognostic systems are crucial for clinical management to assess disease severity and predict outcomes. The primary objective of this study was to assess the epidemiological characteristics and clinical outcomes of patients with Stevens-Johnson syndrome (SJS), toxic epidermal necrolysis (TEN), and SJS/TEN overlap over a 17-year period at a specialized burn center. The secondary objectives were to evaluate the performance of existing prognostic scoring systems (SCORTEN, Re-SCORTEN, and ABCD-10) in predicting mortality and to propose a novel classification tree model to improve mortality prediction. Materials and Methods: A 17-year retrospective study at a burn center included 68 patients with SJS, SJS/TEN overlap, or TEN. Demographic, clinical, laboratory data, and prognostic scores (SCORTEN, Re-SCORTEN, ABCD-10) were collected and analyzed for associations with mortality. A classification tree was created to detect unknown determinants of SJS/TEN mortality. Results: The drug most frequently associated with the occurrence of SJS/TEN was metamizole. The mortality rate was 51%. Affected body surface area, platelet count, and serum blood urea nitrogen differed significantly between survivors and non-survivors. Regarding the scoring systems, only the Re-SCORTEN showed reliable differentiation for these groups. A classification tree model achieved an accuracy of 89% in predicting the mortality risk. In the ROC curve analysis, the AUC values were 0.88 for the classification tree, 0.66 for Re-SCORTEN, 0.61 for SCORTEN, and 0.56 for ABCD-10. Conclusions: This study explores mortality predictors in SJS/TEN via a classification tree model, highlighting potential factors for further investigation. While cautioning against immediate clinical application due to data constraints, the findings underscore the need for larger studies to validate and refine prediction models in this context.

Impact of urgent care centers on emergency department visits in Al Madina Al Munawara: A pre-post study.

To evaluate the impact of UCCs on reducing non-urgent Emergency Department (ED) visits and improving patient flow, focusing on metrics such as door-to-doctor time, doctor-to-decision time, and overall patient disposition. This observational cohort pre-post study analyzed data from 198,050 ED visits to King Fahad Hospital, Al Madina Al Munawara between June 2021 and May 2023 and compared visit patterns before and after UCC implementation. Post-UCC implementation, the average door-to-doctor time decreased but was not statistically significant. Significant reductions were observed in doctor-to-decision and door-to-disposition times for CTAS 3 patients. However, overall patient flow improvements were not fully realized, highlighting the need for enhanced public awareness and integration of UCCs with EDs. The study shows that while urgent care centers in Al Madina Al Munawara have improved efficiency for some patient categories, they don't fully achieve expected reductions in waiting times and patient flow. Seasonal variations, limited patient awareness, and data constraints affect outcomes.

Ferry: Toward Better Understanding of Input/Output Space for Data Wrangling Scripts.

Understanding the input and output of data wrangling scripts is crucial for various tasks like debugging code and onboarding new data. However, existing research on script understanding primarily focuses on revealing the process of data transformations, lacking the ability to analyze the potential scope, i.e., the space of script inputs and outputs. Meanwhile, constructing input/output space during script analysis is challenging, as the wrangling scripts could be semantically complex and diverse, and the association between different data objects is intricate. To facilitate data workers in understanding the input and output space of wrangling scripts, we summarize ten types of constraints to express table space and build a mapping between data transformations and these constraints to guide the construction of the input/output for individual transformations. Then, we propose a constraint generation model for integrating table constraints across multiple transformations. Based on the model, we develop Ferry, an interactive system that extracts and visualizes the data constraints describing the input and output space of data wrangling scripts, thereby enabling users to grasp the high-level semantics of complex scripts and locate the origins of faulty data transformations. Besides, Ferry provides example input and output data to assist users in interpreting the extracted constraints and checking and resolving the conflicts between these constraints and any uploaded dataset. Ferry's effectiveness and usability are evaluated through two usage scenarios and two case studies, including understanding, debugging, and checking both single and multiple scripts, with and without executable data. Furthermore, an illustrative application is presented to demonstrate Ferry's flexibility.

Glacier thickness modelling and monitoring with geophysical data constraints: A case study on the Indren Glacier (NW Italy)

AbstractThe ongoing global temperature increase has accelerated the mass loss of glaciers worldwide, with Italian alpine glaciers being particularly vulnerable due to their small size, complex geometries and exposition that implies a fast reaction to thermal and hydrological modifications. In such a frame, the Indren Glacier (Aosta Valley, north‐western Italian Alps) provides a valid test site to check the thickness evolution over the last two decades (1999–2020), through an integrated approach combining historical data, on‐site geophysical measurements, remote sensing surveys, modelling and temperature analysis. Using a 2018 helicopter‐based photogrammetric survey and Ground Penetrating Radar (GPR) survey campaigns of 2020, we obtained new input data and constraints to build up an updated thickness model for the whole glacier through the Glacier Thickness Estimation algorithm (GlaTE). Ice thickness is indeed a key parameter to estimate the ice volume and use it as further input in evolutionary models forecasting future scenarios. As a part of this integrated approach, we also analysed remote sensing and temperature data, finding a major modification in the glacier conditions over the last decade. Further comparing these results with previous studies, we identified a significant decrease in ice thickness, and we confirmed the presence of an over‐deepening in the glacier central widest part. This integrated methodology enhances our understanding of glacier dynamics and improves predictions of future changes, offering crucial insights for managing water resources and mitigating natural hazards in the alpine region.

Optimizing county-level infectious respiratory disease forecasts: a pandemic case study integrating social media-based physical and social connectivity networks

ABSTRACT Forecasting infectious respiratory diseases is crucial for effective prevention and intervention strategies. However, existing time series forecasting models that incorporate human mobility data have faced challenges in making localized predictions on a large scale across the country due to data costs and constraints. Using the COVID-19 pandemic as a case study, this research explores whether integrating social media-based place and social connectivity networks can improve predictions of disease transmission at the county level across various regions. Place connectivity networks, derived from Twitter users and tweets, and social connectivity networks, based on Facebook interactions, were used to map spatial and social linkages between locations. These networks were integrated into weekly COVID-19 incidence data across 2,927 U.S. counties using Long Short-Term Memory (LSTM) models. The combined connectivity-weighted model significantly enhanced prediction accuracy, reducing Mean Absolute Percentage Error (MAPE) by 49.38% across 96.62% of the counties, with the greatest improvements observed in urban and Northeastern counties. The results demonstrate that combining connectivity networks enhances prediction accuracy, offering a scalable and sustainable solution for localized disease forecasting on a large scale across diverse geographic areas using publicly accessible social media data.

Trade - Growth Nexus in Nepal: A Gravity Model Approach

Purpose of the Study: This study investigates the relationship between trade and economic growth in Nepal, focusing on how factors such as GDP, population, and geographical distance affect Nepal’s exports and imports. The research question addresses: "How do these factors influence Nepal's trade dynamics?" Methods/Design: A gravity model framework was used, with panel data analysis employing fixed and random effects models. The Hausman test determined the appropriate model. Data on trade partners' GDP, population, and geographical distance were collected to estimate the model. Findings: The analysis shows that Nepal’s exports are positively influenced by the GDP and population of trade partners, while geographical distance negatively impacts trade. Closer countries are more favorable trading partners. The random effects model proved to be the most appropriate estimator, with both the export and import models demonstrating strong explanatory power. Conclusion/Implications: The findings suggest that Nepal should focus on export diversification and strengthen regional trade relations to foster economic growth. Policymakers can use these insights to improve trade strategies and regional cooperation. Limitations of the Study: The study is limited by data constraints and does not account for informal trade flows, which could affect the results. Originality of the Study: This study uniquely applies the gravity model to Nepal’s trade context, addressing challenges faced by landlocked developing countries.

Sustainability and Sociability: Shared Practices and Community Cohesion in U.S. Cohousing

This study argues that environmental-sustainability and adjacent practices and identities are conducive to community cohesion in cohousing in the U.S. The theoretical framework draws from cohousing research based on social practice theory and on degrowth concepts. Data came from a Cohousing Research Network combined individual resident- and community-level survey data set (level-1 N = 272, level-2 N = 39). Two multi-level models of community connection among residents included individual-level factors measuring participation in cohousing practices, community-level measures of green identity, and an individual-level attitudinal indicator of potential discord stemming from perceptions of other residents’ insufficient or excessive commitment to sustainability. Keeping in mind data constraints, results from multilevel linear regression analysis were tentatively supportive. The findings suggest that residents indeed connected to their communities through participation in shared practices and through shared green identities. Further confirmatory and exploratory research is needed.

An Enhanced Gas Sensor Data Classification Method Using Principal Component Analysis and Synthetic Minority Over-Sampling Technique Algorithms.

This study addresses the challenge of multi-dimensional and small gas sensor data classification using a gelatin-carbon black (CB-GE) composite film sensor, achieving 91.7% accuracy in differentiating gas types (ethanol, acetone, and air). Key techniques include Principal Component Analysis (PCA) for dimensionality reduction, the Synthetic Minority Over-sampling Technique (SMOTE) for data augmentation, and the Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) algorithms for classification. PCA improved KNN and SVM classification, boosting the Area Under the Curve (AUC) scores by 15.7% and 25.2%, respectively. SMOTE increased KNN's accuracy by 2.1%, preserving data structure better than polynomial fitting. The results demonstrate a scalable approach to enhancing classification accuracy under data constraints. This approach shows promise for expanding gas sensor applicability in fields where data limitations previously restricted reliability and effectiveness.

A hybrid deep learning-based approach for optimal genotype by environment selection.

The ability to accurately predict the yields of different crop genotypes in response to weather variability is crucial for developing climate resilient crop cultivars. Genotype-environment interactions introduce large variations in crop-climate responses, and are hard to factor in to breeding programs. Data-driven approaches, particularly those based on machine learning, can help guide breeding efforts by factoring in genotype-environment interactions when making yield predictions. Using a new yield dataset containing 93,028 records of soybean hybrids across 159 locations, 28 states, and 13 years, with 5,838 distinct genotypes and daily weather data over a 214-day growing season, we developed two convolutional neural network (CNN) models: one that integrates CNN and fully-connected neural networks (CNN model), and another that incorporates a long short-term memory (LSTM) layer after the CNN component (CNN-LSTM model). By applying the Generalized Ensemble Method (GEM), we combined the CNN-based models and optimized their weights to improve overall predictive performance. The dataset provided unique genotype information on seeds, enabling an investigation into the potential of planting different genotypes based on weather variables. We employed the proposed GEM model to identify the best-performing genotypes across various locations and weather conditions, making yield predictions for all potential genotypes in each specific setting. To assess the performance of the GEM model, we evaluated it on unseen genotype-location combinations, simulating real-world scenarios where new genotypes are introduced. By combining the base models, the GEM ensemble approach provided much better prediction accuracy compared to using the CNN-LSTM model alone and slightly better accuracy than the CNN model, as measured by both RMSE and MAE on the validation and test sets. The proposed data-driven approach can be valuable for genotype selection in scenarios with limited testing years. In addition, we explored the impact of incorporating state-level soil data alongside the weather, location, genotype and year variables. Due to data constraints, including the absence of latitude and longitude details, we used uniform soil variables for all locations within the same state. This limitation restricted our spatial information to state-level knowledge. Our findings suggested that integrating state-level soil variables did not substantially enhance the predictive capabilities of the models. We also performed a feature importance analysis using RMSE change to identify crucial predictors. Location showed the highest RMSE change, followed by genotype and year. Among weather variables, maximum direct normal irradiance (MDNI) and average precipitation (AP) displayed higher RMSE changes, indicating their importance.

Sustainable Economic Growth And Industry Insights: Nonlinear Interaction Analysis of GDP Vs FDI in Malaysia Using SCILAB Programming

Objective: To investigate the nonlinear relationship between Gross Domestic Product (GDP) and Foreign Direct Investment (FDI) in Malaysia, with the aim of providing insights into their bidirectional interactions. Theoretical Framework: The main concepts and theories that underpin the research are nonlinear regression techniques and economic growth models. These frameworks provide a solid basis on understanding the dynamic interaction between GDP and FDI in the context of Malaysia’s economic. Method: The methodology comprises writing Scilab coding to analyze nonlinear regression models. Malaysia's economic data on GDP and FDI were utilized as inputs for the analysis. The study involves modeling the dynamics between these indicators and evaluating their relationship over time. Results and Discussion: The results obtained revealed a significant nonlinear relationship between GDP and FDI in Malaysia. These results are contextualized within the theoretical framework, highlighting the bidirectional nature of the relationship. Possible discrepancies and limitations, including data constraints and assumptions of the nonlinear models, are also considered. Research Implications: This research offers valuable insights for policymakers, helping shape economic planning,and investment strategies. It highlights the importance of FDI in promoting sustainable economic growth and industrial development towards United Nation SDG 8 and SDG 9. Originality/Value: This study contributes by using nonlinear regression techniques and Scilab programming to analyze the complex relationship between GDP and FDI, an approach not widely applied to Malaysia's context. The relevance and value of this research are evidenced by its potential impact on economic policy and its contributions to a deeper understanding of Malaysia's economic development.

Modified Characteristics of Hadronic Interactions in Ultra-High-Energy Cosmic-Ray Showers

Data from multiple experiments suggest that the current interaction models used in Monte Carlo simulations do not correctly reproduce the hadronic interactions in air showers produced by ultra-high-energy cosmic rays (UHECR), in particular – but not limited to – the production of muons during the showers. We have created a large library of UHECR simulations where the interactions at the highest energies are slightly modified in various ways – but always within the constraints of the accelerator data, without any abrupt changes with energy and without assuming any specific mechanism or dramatically new physics at the ultra-high energies. We will find that even when very different properties – cross section, elasticity, and multiplicity – of the interactions are modified, the resulting changes in some air-shower observables are still mutually correlated. Thus, not all possible combinations of changes of observables are easily reproduced by some combination of the modifications. Most prominently, the recent results of the Pierre Auger Observatory, which call for a change in the prediction of both the muon content at ground and the depth of the maximum of longitudinal development of the showers, are rather difficult to reproduce with such modifications, in particular when accounting for other cosmic-ray data. While some of these results are related to the assumptions we place on the modifications, the overall lessons are general and provide valuable insight into how the UHECR data can be interpreted from the point of view of hadronic physics.

Subjective Assessment of Rdoc-Related Constructs in Addiction and Compulsive Disorders: A Scoping Review.

Obsessive-compulsive and related disorders (OCRDs) and disorders due to addictive behavior (DABs) are prevalent conditions that share common neurobiological and behavioral characteristics. This scoping review aims to identify and map the range of subjective assessment tools (e.g., interviews and self-report instruments) for assessing Research Domain Criteria (RDoC) and related constructs underlying DABs and OCRDs, such as impaired response inhibition, habit formation, and compulsivity. A scoping review was performed following the PRISMA Extension for Scoping Reviews (PRISMA ScR). The search was conducted in Medline, Web of Science, PsycINFO, and EMBASE databases. No constraints of data or document type were adopted. The search yielded 615 instruments, of which 79 were deemed transdiagnostic and capable of assessing at least one of the target constructs. Noteworthy tools included are the Cambridge-Chicago Compulsivity Trait Scale, the Self-report Habit Index, the Reward Probability Index, the Threat-Related Reassurance-Seeking Scale, and the recent Positive Valence Systems Scale. Despite the growing body of literature on OCRDs and DABs as new diagnostic chapters and the number of tools with the potential to assess their transdiagnostic constructs clinically, most instruments were designed to capture dimensions or psychopathology not directly (or primarily) related to OCRDs, DABs and the concepts of impaired response inhibition, habit formation, and compulsivity in the context of these conditions. Further studies exploring the correlation between subjective assessments, corresponding behavior paradigms, and neuroimaging data would be of great value in the translation of RDoC constructs and domains into clinical settings.

Modeling performance of data collection systems for high-energy physics

Exponential increases in scientific experimental data are outpacing silicon technology progress, necessitating heterogeneous computing systems—particularly those utilizing machine learning (ML)—to meet future scientific computing demands. The growing importance and complexity of heterogeneous computing systems require systematic modeling to understand and predict the effective roles for ML. We present a model that addresses this need by framing the key aspects of data collection pipelines and constraints and combining them with the important vectors of technology that shape alternatives, computing metrics that allow complex alternatives to be compared. For instance, a data collection pipeline may be characterized by parameters such as sensor sampling rates and the overall relevancy of retrieved samples. Alternatives to this pipeline are enabled by development vectors including ML, parallelization, advancing CMOS, and neuromorphic computing. By calculating metrics for each alternative such as overall F1 score, power, hardware cost, and energy expended per relevant sample, our model allows alternative data collection systems to be rigorously compared. We apply this model to the Compact Muon Solenoid experiment and its planned high luminosity-large hadron collider upgrade, evaluating novel technologies for the data acquisition system (DAQ), including ML-based filtering and parallelized software. The results demonstrate that improvements to early DAQ stages significantly reduce resources required later, with a power reduction of 60% and increased relevant data retrieval per unit power (from 0.065 to 0.31 samples/kJ). However, we predict that further advances will be required in order to meet overall power and cost constraints for the DAQ.

Graph convolutional network for compositional data

Graph convolutional network (GCN) has garnered significant attention and become a powerful tool for learning graph representations. However, when dealing with compositional data prevalent in various fields, the traditional GCN faces theoretical challenges due to the intrinsic constraints of such data. This paper generalizes the spectral graph theory in simplex space, aiming to address the graph structures among observations for compositional data analysis and to extend GCN by assigning mathematical objects of compositions to each vertex of a graph. We propose the graph Fourier transformation in simplex space, based on which a compositional graph convolutional network (CGCN) layer is introduced. This novel layer enables a GCN to appropriately capture the sample space of compositional data, allowing it to handle compositional features as model inputs. We then propose a new GCN architecture called COMP-GCN, incorporating the CGCN layer at the initial stage. We evaluate the effectiveness of COMP-GCN through simulation studies and two real-world applications: stock networks derived from co-investors in the Chinese stock market and student social networks based on co-locations in campus activities. The results demonstrate its superior performance over competitive methods with modest additional computational cost compared to traditional GCN. Our findings suggest the potential of the proposed model to inspire a new class of powerful algorithms for graph inference on compositional data in virtue of the generalization of graph convolution on simplex space.

Exploiting a non-mainstream financial scheme to innovate: SMEs in the developing world

PurposeThe study aims to explore the role of non-mainstream financial schemes in supporting innovation within SMEs in developing countries, particularly in sub-Saharan Africa. It investigates how informal credit, business group affiliation and foreign and state ownership arrangements influence SMEs’ innovative activities in environments with limited access to formal financial resources.Design/methodology/approachThe research utilizes data from the World Bank’s Enterprise Surveys, focusing on 8,466 firms across 11 sub-Saharan African countries from 2011 to 2020. A logistic regression analysis was conducted to assess the impact of various financial sources on SMEs’ innovation outputs, particularly incremental innovations, due to data constraints on radical innovations.FindingsThe findings reveal that informal credit significantly supports SME innovation, while business group resources can hinder innovative activities by restricting firms to routine tasks. State ownership positively influences innovation, whereas the impact of foreign ownership is inconclusive. These results highlight the critical role of alternative financial mechanisms in the innovation activities of SMEs in resource-limited settings.Originality/valueThis study contributes to the literature by providing empirical evidence on the effects of non-mainstream financial schemes on SME innovation in developing countries. It offers new theoretical insights into how SMEs navigate financial constraints to foster innovation and suggests policy implications for improving financial support systems for SMEs in such contexts. The research underscores the importance of contextualizing entrepreneurship studies to better understand the unique challenges and opportunities faced by SMEs in developing regions.

Automatic Reconstruction of Reservoir Geological Bodies Based on Improved Conditioning Spectral Normalization Generative Adversarial Network

For reservoir structural models with obvious nonstationary and heterogeneous characteristics, traditional geostatistical simulation methods tend to produce suboptimal results. Additionally, these methods are computationally resource-intensive in consecutive simulation processes. Thanks to the feature extraction capability of deep learning, the generative adversarial network-based method can overcome the limitations of geostatistical simulation and effectively portray the structural attributes of the reservoir models. However, the fixed receptive fields may restrict the extraction of local geospatial multiscale features, while the gradient anomalies and mode collapse during the training process can cause poor reconstruction. Moreover, the sparsely distributed conditioning data lead to possible noise and artifacts in the simulation results due to its weak constraint ability. Therefore, this paper proposes an improved conditioning spectral normalization generation adversarial network framework (CSNGAN-ASPP) to achieve efficient and automatic reconstruction of reservoir geological bodies under sparse hard data constraints. Specifically, CSNGAN-ASPP features an encoder-decoder type generator with an atrous spatial pyramid pooling (ASPP) structure, which effectively identifies and extracts multi-scale geological features. A spectral normalization strategy is integrated into the discriminator to enhance the network stability. Attention mechanisms are incorporated to focus on the critical features. In addition, a joint loss function is defined to optimize the network parameters and thereby ensure the realism and accuracy of the simulation results. Three types of reservoir model were introduced to validate the reconstruction performance of CSNGAN-ASPP. The results show that they not only accurately conform to conditioning data constraints but also closely match the reference model in terms of spatial variance, channel connectivity, and facies attribute distribution. For the trained CSNGAN-ASPP, multiple corresponding simulation results can be obtained quickly through inputting conditioning data, thus achieving efficient and automatic reservoir geological model reconstruction.