Data-driven Machine Learning Approach Research Articles

Machine Learning-Assisted Hartree–Fock Approach for Energy Level Calculations in the Neutral Ytterbium Atom

Data-driven machine learning approaches with precise predictive capabilities are proposed to address the long-standing challenges in the calculation of complex many-electron atomic systems, including high computational costs and limited accuracy. In this work, we develop a general workflow for machine learning-assisted atomic structure calculations based on the Cowan code’s Hartree–Fock with relativistic corrections (HFR) theory. The workflow incorporates enhanced ElasticNet and XGBoost algorithms, refined using entropy weight methodology to optimize performance. This semi-empirical framework is applied to calculate and analyze the excited state energy levels of the 4f closed-shell Yb I atom, providing insights into the applicability of different algorithms under various conditions. The reliability and advantages of this innovative approach are demonstrated through comprehensive comparisons with ab initio calculations, experimental data, and other theoretical results.

Revolutionizing inventory management: A comprehensive automated data-driven model using power BI incorporating industry 4.0

This is a study on advanced inventory management approaches in the mechanical industry using data-driven models, accompanied by Power BI analytics. Drawing insights from broad collaboration with relevant industrial expertise at different organizational scales, our research work addresses the intricacies and challenges in maintaining optimal inventory levels concerning growing industries and operational risks. In this global mechanical firm, we analyzed historical data using descriptive and predictive analytics to understand the trends of the past and to forecast future inventory needs. Descriptive analytics provided base insights about the present status of inventory, while predictive analytics helped proactive management of stock levels classified into on-hand and critical level categories for greater efficiency in strategies related to planning at the coordination end with its suppliers and mitigating risks. Core to our methodology will be the implementation of an automated data-driven machine-learning approach that brings minimal intervention from humans to solve some of the most critical problems in inventory management. One major contribution this study adds is a major product: a Power BI dashboard that visualizes critical part numbers falling below threshold values and recognizes suppliers with historical shipment shortages. Armed with this intuitive dashboard, supplier-coordinating engineers are better placed to take proactive action on inventory shortages, greatly improving operational efficiency in supply chain resilience.

Modeling of the metal–insulator transition temperature in alio-valently doped VO2 through symbolic regression

The correlated semiconductor vanadium dioxide (VO2) exhibits an insulator–metal transition (IMT) near room temperature, which is of interest in various device applications. Precise IMT temperature control is crucial to determine the use cases across technologies such as thermochromic windows, actuators for robots or neuronal oscillators. Doping the cation or anion sites can modulate the IMT by several tens of degrees and control hysteresis. However, modeling the effects of control parameters (e.g., doping concentration, type of dopants) is challenging due to complex experimental procedures and limited data, hindering the use of traditional data-driven machine learning approaches. Symbolic regression (SR) can bridge this gap by identifying nonlinear expressions connecting key input parameters to target properties, even with small data sets. In this work, we develop SR models to capture the IMT trends in VO2 influenced by different dopant parameters. Using experimental data from the literature, our study reveals a dual nature of the IMT temperature with varying tungsten (W) doping concentrations. The symbolic model captures data trends and accounts for experimental variability, providing a complementary approach to first-principles calculations. Our feature-driven analysis across a broader class of dopants informs selectivity and provides qualitative insights into tuning phase transition properties valuable for neuromorphic computing and thermochromic windows.

Data-driven machine learning approaches for precise lithofacies identification in complex geological environments

ABSTRACT Reservoir characterization is a vital task within the oil and gas industry, with the identification of lithofacies in subsurface formations being a fundamental aspect of this process. However, lithofacies identification in complex geological environments with high dimensions, such as the Lower Indus Basin in Pakistan, poses a notable challenge, especially when dealing with limited data. To address this issue, we propose four common data-driven machine learning approaches: multi-resolution graph-based clustering (MRGC), artificial neural networks (ANN), K-nearest neighbors (KNN), and self-organizing map (SOM). We utilized these proposed approaches to assess their performance in scenarios with varying core sample availability, specifically evaluating their effectiveness in identifying lithofacies within the Lower Goru formation of the middle Indus Basin. The study reveals that in scenarios with a limited number of core samples, MRGC is the preferred choice, while KNN or MRGC is more suitable for larger datasets. The results demonstrate the superior performance of MRGC and KNN in lithofacies identification within the specified geological environment, with SOM following closely behind, and ANN exhibiting comparatively lower efficacy. The accurate identification of lithofacies from the selected model is complemented by the application of the truncated Gaussian simulation method for facies modeling. Comparative results confirm the excellent agreement between the model identification of lithofacies from well logs and electro-facies obtained from the truncated Gaussian simulation electro-facies volume. This study highlights the crucial role of selecting the right machine learning approach for precise lithofacies identification and modeling in complex geological environments. The comparative analysis provides practitioners in the petroleum industry with insights into the strengths and limitations of each method, enhancing existing knowledge. In conclusion, this research emphasizes the significance of comprehensive research and method selection for advancing lithofacies identification in diverse formations or study areas, ultimately benefiting the broader field of subsurface characterization in the petroleum industry.

Prediction of Social Engagement in Long-Term Care Homes by Sex: A Population-Based Analysis Using Machine Learning.

The objective of this study was to use population-based clinical assessment data to build and evaluate machine-learning models for predicting social engagement among female and male residents of long-term care (LTC) homes. Routine clinical assessments from 203,970 unique residents in 647 LTC homes in Ontario, Canada, collected between April 1, 2010, and March 31, 2020, were used to build predictive models for the Index of Social Engagement (ISE) using a data-driven machine-learning approach. General and sex-specific models were built to predict the ISE. The models showed a moderate prediction ability, with random forest emerging as the optimal model. Mean absolute errors were 0.71 and 0.73 in females and males, respectively, using general models and 0.69 and 0.73 using sex-specific models. Variables most highly correlated with the ISE, including activity pursuits, cognition, and physical health and functioning, differed little by sex. Factors associated with social engagement were similar in female and male residents.

Machine learning analysis of a large set of homopolymers to predict glass transition temperatures

Glass transition temperature of polymers, Tg, is an important thermophysical property, which sometimes can be difficult to measure experimentally. In this regard, data-driven machine learning approaches are important alternatives to assess Tg values, in a high-throughput way. In this study, a large dataset of more than 900 polymers with reported glass transition temperature (Tg) was assembled from various public sources in order to develop a predictive model depicting the structure-property relationships. The collected dataset was curated, explored via cluster analysis, and then split into training and test sets for validation purposes and then polymer structures characterized by molecular descriptors. To find the models, several machine learning techniques, including multiple linear regression (MLR), k-nearest neighbor (k-NN), support vector machine (SVM), random forest (RF), gaussian processes for regression (GPR), and multi-layer perceptron (MLP) were explored. As result, a model with the subset of 15 descriptors accurately predicting the glass transition temperatures was developed. The electronic effect indices were determined to be important properties that positively contribute to the Tg values. The SVM-based model showed the best performance with determination coefficients (R2) of 0.813 and 0.770, for training and test sets, respectively. Also, the SVM model showed the lowest estimation error, RMSE = 0.062. In addition, the developed structure-property model was implemented as a web app to be used as an online computational tool to design and evaluate new homopolymers with desired glass transition profiles.

Data-driven machine learning approaches for simultaneous prediction of peak particle velocity and frequency induced by rock blasting in mining

The vibrations generated by rock blasting are a serious and hazardous outcome of these activities, causing harmful effects on the surrounding environment as well as the nearby residents. Both the local ecology and human communities suffer from the consequences of these vibrations. Assessing the severity of blasting vibrations necessitates a thorough evaluation of Peak Particle Velocity (PPV) and frequency, which are essential parameters for measuring vibration velocity. Accurate prediction of vibration occurrence is critically important. Therefore, this study employs five machine learning models for predicting the PPV and frequency resulting from quarry blasting. This work compares five machine learning models (XGBoost, Catboost, Bagging, Gradient Boosting, and Random Forest Regression) to choose the most efficient performance model. The performance evaluation of each five machine learning models demonstrates each model achieved a performance of more than 0.90 during the testing phase, there was a strong correlation observed between the actual and the predicted ones. The analysis of performance metrics shows Catboost regression model demonstrate better performance prediction comparing with the other models with R2 = 0.983, MSE = 0.000078, RMSE = 0.008, NRMSE = 0.019, MAD = 0.004, MAPE= 35.197 in the PPV prediction, and R2 = 0.975, MSE = 0.000243, RMSE = 0.015, NRMSE = 0.031, MAD = 0.008, MAPE = 37.281 for the frequency prediction. This study will help mining engineers and blasting experts to select the best machine learning model and its hyperparameters in estimating ground vibration, and frequency. In the context of the mining and civil industry, the application of this study offers significant potential for enhancing safety protocols and optimizing operational efficiency. By employing machine learning models, this research aims to accurately predict and assess ground vibrations with frequency resulting from rock blasting.

Machine learning based predictive analysis of DNA cleavage induced by diverse nanomaterials

DNA cleavage by nanomaterials has the potential to be utilized as an innovative tool for gene editing. Numerous nanomaterials exhibiting DNA cleavage properties have been identified and cataloged. Yet, the exploitation of property data through data-driven machine-learning approaches remains unexplored. A database was developed, compiling thirty distinctive characteristics, encompassing physical and chemical properties, as well as experimental conditions of nanomaterials that have demonstrated DNA cleavage capability such as in articles published over the past two decades. The DNA cleavage effect and efficiency of nanomaterials were predicted using machine learning algorithms such as support vector machines, deep neural networks, and random forest, and a classification accuracy of 0.93 for the cleavage effect was achieved. Moreover, the potential of utilizing larger datasets to enhance the predictive capacity of models was discussed. The findings indicate the feasibility of predicting nanomaterial properties based on experimental data. Evaluating the performance and effectiveness of the machine learning models trained using the existing data can furnish valuable insights for future materials research endeavors, especially for the design of DNA cleavage with specific sites.

Enhancing box-wing design efficiency through machine learning based optimization

The optimization of wings typically relies on computationally intensive high-fidelity simulations, which restrict the quick exploration of design spaces. To address this problem, this paper introduces a methodology dedicated to optimizing box wing configurations using low-fidelity data driven machine learning approach. This technique showcases its practicality through the utilization of a tailored low-fidelity machine learning technique, specifically designed for early-stage wing configuration. By employing surrogate model trained on small dataset derived from low-fidelity simulations, our method aims to predict outputs within an acceptable range. This strategy significantly mitigates computational costs and expedites the design exploration process. The methodology’s validation relies on its successful application in optimizing the box wing of PARSIFAL, serving as a benchmark, while the primary focus remains on optimizing the newly designed box wing by Bionica. Applying this method to the Bionica configuration led to a notable 14% improvement in overall aerodynamic efficiency. Furthermore, all the optimized results obtained from machine learning model undergo rigorous assessments through the high-fidelity RANS analysis for confirmation. This methodology introduces innovative approach that aims to streamline computational processes, potentially reducing the time and resources required compared to traditional optimization methods.

Advancing container port traffic simulation: A data-driven machine learning approach in sparse data environments

Efficient truck dispatching strategies are paramount in container terminal operations. The quality of these strategies heavily relies on accurate and expedient simulations, which provide a crucial platform for training and evaluating dispatching algorithms. In this study, we introduce data-driven machine learning methods to enhance container port truck dispatching simulation accuracy. These methods effectively surrogate the intersections within the simulation, thereby increasing the accuracy of simulated outcomes without imposing significant computational overhead in sparse data environments. We incorporate three data-driven learning methods: genetic programming (GP), reinforcement learning (RL), and a GP and RL hybrid heuristic (GPRL-H) approach. The GPRL-H method proved the most efficacious through a detailed comparative study, striking an effective balance between simulation accuracy and computational efficiency. It reduced the error rate of simulation from approximately 35% to about 7%, while also halving the simulation time compared to the RL-based method. Our proposed method also does not rely on precise Global Positioning System (GPS) data to simulate truck operations within a port accurately. Demonstrating robustness and adaptability, this approach holds promise for extending beyond port operations to improve the simulation accuracy of vehicle operations in various scenarios characterized by sparse data.

Organizational and Individual Contributing Factors to Safety Climate in Healthcare Industries - Bayesian Network Predictive Modeling Approach.

The current study aims to identify individual and joint drivers that significantly influence the safety climate in healthcare industries by using Bayesian Network simulations for an in-depth analysis. Survey data were collected from 452 employees from two branches of one hospital in China for a study about workplace safety. The original English surveys were translated into Chinese using the back-translation procedure recommended by Brislin. Employees were asked to complete two online surveys with one month in between each administration. The sample was 42% doctors and 58% nurses. A Bayesian Network (BN) model, based on theory, was updated and complemented with expert knowledge. A graphical model based on expert knowledge and data-driven machine learning approaches was used to refine the BN structure, representing interrelationships among all studied variables. The BN model was employed to identify the best key drivers and joint strategies for safety climate improvement. The Bayesian Network model demonstrated a good overall fit. The Euclidean distance metric was used to assess the influence between connected variables, with interpersonal trust and locus of control having the strongest independent effects on safety climate among the five contributing factors. Joint strategies, particularly joint optimization of error disclosure culture and interpersonal trust, as well as error disclosure culture and self-efficacy, were most effective in promoting a safe climate. The findings suggest that hospital safety climate can be improved by providing a psychologically safe error disclosure culture and enhancing interpersonal trust among employees and their self-efficacy.

Characterizing the clinical heterogeneity of early symptomatic Alzheimer's disease: a data-driven machine learning approach.

Alzheimer's disease (AD) is highly heterogeneous, with substantial individual variabilities in clinical progression and neurobiology. Amyloid deposition has been thought to drive cognitive decline and thus a major contributor to the variations in cognitive deterioration in AD. However, the clinical heterogeneity of patients with early symptomatic AD (mild cognitive impairment or mild dementia due to AD) already with evidence of amyloid abnormality in the brain is still unknown. Participants with a baseline diagnosis of mild cognitive impairment or mild dementia, a positive amyloid-PET scan, and more than one follow-up Alzheimer's Disease Assessment Scale-Cognitive Subscale-13 (ADAS-Cog-13) administration within a period of 5-year follow-up were selected from the Alzheimer's Disease Neuroimaging Initiative database (n = 421; age = 73±7; years of education = 16 ± 3; percentage of female gender = 43%; distribution of APOE4 carriers = 68%). A non-parametric k-means longitudinal clustering analysis in the context of the ADAS-Cog-13 data was performed to identify cognitive subtypes. We found a highly variable profile of cognitive decline among patients with early AD and identified 4 clusters characterized by distinct rates of cognitive progression. Among the groups there were significant differences in the magnitude of rates of changes in other cognitive and functional outcomes, clinical progression from mild cognitive impairment to dementia, and changes in markers presumed to reflect neurodegeneration and neuronal injury. A nomogram based on a simplified logistic regression model predicted steep cognitive trajectory with an AUC of 0.912 (95% CI: 0.88 - 0.94). Simulation of clinical trials suggested that the incorporation of the nomogram into enrichment strategies would reduce the required sample sizes from 926.8 (95% CI: 822.6 - 1057.5) to 400.9 (95% CI: 306.9 - 516.8). Our findings show usefulness in the stratification of patients in early AD and may thus increase the chances of finding a treatment for future AD clinical trials.

Modeling the thermal transport properties of hydrogen and its mixtures with greenhouse gas impurities: A data-driven machine learning approach

This study introduces machine learning (ML) algorithms to predict hydrogen (H2) thermodynamic properties for geological storage, focusing on its mixtures with natural gas, nitrogen (N2), and carbon dioxide (CO2). Employing 1167 data samples, this research utilizes multiple linear regression (MLR), random forest (RF), gradient boosting (GB), and decision tree (DT) models, enhanced by sensitivity analysis for feature engineering. GB model's superior accuracy and efficiency over conventional statistical linear regression methods. The finding reveals that the GB model achieves superior performance, with an R2 value of 0.9998 for thermal capacity and the lowest mean absolute error (MAE)/root mean square error (RMSE) across all H2 properties—density (1.51%, 2.06 kg/m3), viscosity (0.000078%, 0.000117 cp), thermal conductivity (0.000427%, 0.000849 W/m.K), and thermal capacity (8.79%, 26.06 J/g.K). Moreover, this ML-based approach not only demonstrates remarkable accuracy and efficiency but also suggests the potential of smart paradigms to further enhance the safety and transportability of underground hydrogen storage projects. Ultimately, this work will help the reservoir simulation at field scale to be more robust and reliable.

Ice Phase Classification Made Easy with Score-Based Denoising.

Accurate identification of ice phases is essential for understanding various physicochemical phenomena. However, such classification for structures simulated with molecular dynamics is complicated by the complex symmetries of ice polymorphs and thermal fluctuations. For this purpose, both traditional order parameters and data-driven machine learning approaches have been employed, but they often rely on expert intuition, specific geometric information, or large training data sets. In this work, we present an unsupervised phase classification framework that combines a score-based denoiser model with a subsequent model-free classification method to accurately identify ice phases. The denoiser model is trained on perturbed synthetic data of ideal reference structures, eliminating the need for large data sets and labeling efforts. The classification step utilizes the smooth overlap of atomic position (SOAP) descriptors as the atomic fingerprint, ensuring Euclidean symmetries and transferability to various structural systems. Our approach achieves a remarkable 100% accuracy in distinguishing ice phases of test trajectories using only seven ideal reference structures of ice phases as model inputs. This demonstrates the generalizability of the score-based denoiser model in facilitating phase identification for complex molecular systems. The proposed classification strategy can be broadly applied to investigate structural evolution and phase identification for a wide range of materials, offering new insights into the fundamental understanding of water and other complex systems.

Large-scale analysis to identify risk factors for ovarian cancer

ObjectiveOvarian cancer is characterized by late-stage diagnoses and poor prognosis. We aimed to identify factors that can inform prevention and early detection of ovarian cancer.MethodsWe used a data-driven machine learning...

Adaptive machine learning surrogate based multiobjective optimization for scavenging residual saltwater behind subsurface dams

High-fidelity physically based groundwater flow and solute transport models have been limited for seawater intrusion remediation design because of computationally intensive evolutionary algorithms. Data-driven machine learning approaches are promising to substitute expensive-cost groundwater numerical models within optimization due to computing efficiency. However, machine learning surrogates may accumulate error of forecasting and thereby result in infeasible optimal solutions. To achieve desired fidelity level, this study proposes a novel adaptive machine learning surrogate based multiobjective optimization method for coastal aquifer desalination. An adaptive modeling algorithm is newly introduced to iteratively retrain poorly-performed machine learning models and enhance predicting accuracy. The method is demonstrated to seek optimal extraction and injection strategies for scavenging residual saltwater trapped in an upstream aquifer behind a subsurface dam. Two conflicting objectives of minimizing the total extraction-injection rate and maximizing saltwater removal effectiveness are considered. Three machine learning models including artificial neural network (ANN), Gaussian process (GP) and response surface regression model (RSR) are developed to replace a high-fidelity seawater intrusion model for predicting chloride concentration and salinity mass. Non-dominated Sorting Genetic Algorithm II (NSGA-II) is employed to derive Pareto fronts. Pareto optimal solutions obtained from machine learning models are compared against those from the seawater intrusion model. Results indicate that the developed machine learning models do not only have strong predicting capability, but also maintain good quality of Pareto optimal solutions, while achieving substantial computational saving up to 95%. Especially, the adaptively retrained RSR model inhibits error accumulation of forecasting and accelerates correct convergence to the true Pareto front. The proposed method is found superior performance in accurate convergence, widely-spread diversity as well as high computing efficiency than conventional methods.

Cross-cultural comparison of beauty judgments in visual art using machine learning analysis of art attribute predictors among Japanese and German speakers

In empirical art research, understanding how viewers judge visual artworks as beautiful is often explored through the study of attributes—specific inherent characteristics or artwork features such as color, complexity, and emotional expressiveness. These attributes form the basis for subjective evaluations, including the judgment of beauty. Building on this conceptual framework, our study examines the beauty judgments of 54 Western artworks made by native Japanese and German speakers, utilizing an extreme randomized trees model—a data-driven machine learning approach—to investigate cross-cultural differences in evaluation behavior. Our analysis of 17 attributes revealed that visual harmony, color variety, valence, and complexity significantly influenced beauty judgments across both cultural cohorts. Notably, preferences for complexity diverged significantly: while the native Japanese speakers found simpler artworks as more beautiful, the native German speakers evaluated more complex artworks as more beautiful. Further cultural distinctions were observed: for the native German speakers, emotional expressiveness was a significant factor, whereas for the native Japanese speakers, attributes such as brushwork, color world, and saturation were more impactful. Our findings illuminate the nuanced role that cultural context plays in shaping aesthetic judgments and demonstrate the utility of machine learning in unravelling these complex dynamics. This research not only advances our understanding of how beauty is judged in visual art—considering self-evaluated attributes—across different cultures but also underscores the potential of machine learning to enhance our comprehension of the aesthetic evaluation of visual artworks.

Influences of the composition design on the microstructure and properties of laser additive manufacturing TiAl alloy coatings

The utilization of titanium aluminide (TiAl) alloys is widespread in the high-temperature sectors of the aerospace industry, despite their challenging manufacturing process and inherent low toughness. Laser additive manufacturing (LAM), as a solution to these challenges, enhances the capability to form complex components and mechanical properties. The influence of process parameters, heat treatment methods, and reinforcing components on the microstructure of TiAl alloys was studied during the LAM process. The principles of designing TiAl alloys were emphasized from the action mechanism of reinforcing components. A classification standard for the component design was proposed based on the different roles of β-stabilizing elements, α-stabilizing elements, rare earth elements, and composites (a combination of the aforementioned components). Furthermore, the work discussed the evolution mechanism of the microstructure and the strengthening mechanism of mechanical properties in TiAl alloys prepared using LAM. These included solid solution strengthening, precipitation strengthening, grain boundary strengthening, grain refinement, and lattice distortion toughening. Finally, future research directions in LAM TiAl alloys were proposed—the innovative design of TiAl alloys through the structural or interfacial control, the development of multiscale characterization techniques for microstructures, and the advancement of high-performance TiAl alloys using data-driven machine learning approaches.

Fabrication of an asymmetric supercapacitor using a novel electrode design and introduce a robust machine learning model for its performance evaluation

Attaining high performance supercapacitors requires comprehensive investigations into electrode and electrolyte properties, along with the recognition of operational parameters. The present research focuses on fabricating an asymmetric two-electrode supercapacitor (ASC) using Co3O4/VS4-rGO@NF and conductive carbon cloth (CC). This work evaluates the ASC performance using experimental and machine learning techniques. Experimental analyses are performed to investigate the influence of morphology, structure, specific surface area, and surface charge on electrochemical performance. The ASC exhibits a Csp of 238.8 F g−1 at 0.5 Ag-1 and demonstrates capacity retention after 8500 charge-discharge cycles, showcasing efficient electrochemical processes. Moreover, the ASC device shows an energy density of 74.6 Wh kg−1 at a power density of 1500 W kg−1. A data-driven machine learning approach based on artificial neural networks (ANN) predicates ASC's capacity with accuracy. The developed model accurately predicts Csp values for different architectures, ranging from 208.2 to 214.0 Fg-1 at 0.5 Ag-1. When comparing experimental and predicted values using the optimized architecture, the relative error averages 5.8 %. The successful results obtained from this study not only introduce a novel electrode design but also propose a robust machine learning model that facilitates accurate decision-making in experimental research.

Machine Learning Approaches for Vehicle Counting on Bridges Based on Global Ground-Based Radar Data

Abstract. This study introduces a novel data-driven approach for classifying and estimating the number of vehicles crossing a bridge solely on non-invasive ground-based radar time series data (GBR data). GBR is used to measure the bridge displacement remotely. It has recently been investigated for remote bridge weigh-in-motion (BWIM). BWIM mainly focuses on single-vehicle events. However, events with several vehicles should be exploited to increase the amount of data. Therefore, extracting the number of involved vehicles in the first step would be beneficial. Acquiring such information from global bridge responses such as displacement can be challenging. This study indicates that a data-driven machine learning approach can extract the vehicle count from GBR time series data. When classifying events according to the number of vehicles, we achieve a balanced accuracy of up to 80 % on an imbalanced dataset. We also try to estimate the number of cars and trucks separately via regression and acquire a R2 of 0.8. Finally, we show the impact of the data augmentation methods we apply to the GBR data to tackle the skew in the dataset using the feature importance of Random Forests.