Flood hazard potential evaluation using decision tree state-of-the-art models.

Floods occur frequently in Romania and throughout the world and are one of the most devastating natural disasters that impact people's lives. Therefore, in order to reduce the potential damages, an accurate identification of surfaces susceptible to flood phenomena is mandatory. In this regard, the quantitative calculation of flood susceptibility has become a very popular practice in the scientific research. With the development of modern computerized methods such as geographic information system and machine learning models, and as a result of the possibility of combining them, the determination of areas susceptible to floods has become increasingly accurate, and the algorithms used are increasingly varied. Some of the most used and highly accurate machine learning algorithms are the decision tree models. Therefore, in the present study focusing on flood susceptibility zonation mapping in the Trotus River basin, the following algorithms were applied: forest by penalizing attribute-weights of evidence (forest-PA-WOE), best first decision tree-WOE, alternating decision tree-WOE, and logistic regression-WOE. The best performant, characterized by a maximum accuracy of 0.981, proved to be forest-PA-WOE, whereas in terms of flood exposure, an area of over 16.22% of the Trotus basin is exposed to high and very high floods susceptibility. The performances applied models in the present work are higher than the models applied in the previous studies in the same study area. Moreover, it should be noted that the accuracy of the models is similar with the accuracies of the decision tree models achieved in the studies focused on other areas across the world. Therefore, we can state that the models applied in the present research can be successfully used in by the researchers in other case studies. The findings of this research may substantially map the flood risk areas and further aid watershed managers in limiting and remediating flood damage in the data-scarce regions. Moreover, the results of this study can be a very useful for the hazard management and planning authorities.

Ensemble models based on radial basis function network for landslide susceptibility mapping.

Ensemble learning techniques have shown promise in improving the accuracy of landslide models by combining multiple models to achieve better predictive performance. In this study, several ensemble methods (Dagging, Bagging, and Decorate) and a radial basis function classifier (RBFC) were combined to predict landslide susceptibility in the Trung Khanh district of the Cao Bang Province, Vietnam. The ensemble models were developed using a geospatial database containing 45 historical landslides (1074 points) and thirteen influencing variables characterizing the topography, geology, land use/cover, and human activities of the study area. The performance of the models was evaluated based on the area under the receiver operating characteristic curve (AUC) and several other performance metrics, including positive predictive value (PPV), negative predictive value (NPV), sensitivity (SST), specificity (SPF), accuracy (ACC), and root mean square error (RMSE). The Bagging-RBFC model with PPV = 86%, NPV = 95%, SST = 95%, SPF = 87%, ACC = 91%, RMSE = 0.297, and AUC = 98% was found to be the most accurate model for the prediction of landslide susceptibility, followed by the Dagging-RBFC, Decorate-RBFC, and single RBFC models. The study demonstrates the efficacy of ensemble learning techniques in developing reliable landslide predictive models, which can ultimately save lives and reduce infrastructure damage in landslide-prone regions worldwide.

Multiplex RT Real-Time PCR Based on Target Failure to Detect and Identify Different Variants of SARS-CoV-2: A Feasible Method That Can Be Applied in Clinical Laboratories.

Shortly after its emergence, Omicron and its sub-variants have quickly replaced the Delta variant during the current COVID-19 outbreaks in Vietnam and around the world. To enable the rapid and timely detection of existing and future variants for epidemiological surveillance and diagnostic applications, a robust, economical real-time PCR method that can specifically and sensitively detect and identify multiple different circulating variants is needed. The principle of target- failure (TF) real-time PCR is simple. If a target contains a deletion mutation, then there is a mismatch with the primer or probe, and the real-time PCR will fail to amplify the target. In this study, we designed and evaluated a novel multiplex RT real-time PCR (MPL RT-rPCR) based on the principle of target failure to detect and identify different variants of SARS-CoV-2 directly from the nasopharyngeal swabs collected from COVID-19 suspected cases. The primers and probes were designed based on the specific deletion mutations of current circulating variants. To evaluate the results from the MPL RT-rPCR, this study also designed nine pairs of primers for amplifying and sequencing of nine fragments from the S gene containing mutations of known variants. We demonstrated that (i) our MPL RT-rPCR was able to accurately detect multiple variants that existed in a single sample; (ii) the limit of detection of the MPL RT-rPCR in the detection of the variants ranged from 1 to 10 copies for Omicron BA.2 and BA.5, and from 10 to 100 copies for Delta, Omicron BA.1, recombination of BA.1 and BA.2, and BA.4; (iii) between January and September 2022, Omicron BA.1 emerged and co-existed with the Delta variant during the early period, both of which were rapidly replaced by Omicron BA.2, and this was followed by Omicron BA.5 as the dominant variant toward the later period. Our results showed that SARS-CoV-2 variants rapidly evolved within a short period of time, proving the importance of a robust, economical, and easy-to-access method not just for epidemiological surveillance but also for diagnoses around the world where SARS-CoV-2 variants remain the WHO's highest health concern. Our highly sensitive and specific MPL RT-rPCR is considered suitable for further implementation in many laboratories, especially in developing countries.

Flood risk mapping and analysis using an integrated framework of machine learning models and analytic hierarchy process.

In this study, a new approach of machine learning (ML) models integrated with the analytic hierarchy process (AHP) method was proposed to develop a holistic flood risk assessment map. Flood susceptibility maps were created using ML techniques. AHP was utilized to combine flood vulnerability and exposure criteria. We selected Quang Binh province of Vietnam as a case study and collected available data, including 696 flooding locations of historical flooding events in 2007, 2010, 2016, and 2020; and flood influencing factors of elevation, slope, curvature, flow direction, flow accumulation, distance from river, river density, land cover, geology, and rainfall. These data were used to construct training and testing datasets. The susceptibility models were validated and compared using statistical techniques. An integrated flood risk assessment framework was proposed to incorporate flood hazard (flood susceptibility), flood exposure (distance from river, land use, population density, and rainfall), and flood vulnerability (poverty rate, number of freshwater stations, road density, number of schools, and healthcare facilities). Model validation suggested that deep learning has the best performance of AUC = 0.984 compared with other ensemble models of MultiBoostAB Ensemble (0.958), Random SubSpace Ensemble (0.962), and credal decision tree (AUC = 0.918). The final flood risk map shows 5075 ha (0.63%) in extremely high risk, 47,955 ha (5.95%) in high-risk, 40,460 ha (5.02%) in medium risk, 431,908 ha (53.55%) in low risk areas, and 281,127 ha (34.86%) in very low risk. The present study highlights that the integration of ML models and AHP is a promising framework for mapping flood risks in flood-prone areas.

Flood susceptibility evaluation through deep learning optimizer ensembles and GIS techniques.

It is difficult to predict and model with an accurate model the floods, that are one of the most destructive risks across the earth's surface. The main objective of this research is to show the prediction power of three ensemble algorithms with respect to flood susceptibility estimation. These algorithms are: Iterative Classifier Optimizer - Alternating Decision Tree - Frequency Ratio (ICO-ADT-FR), Iterative Classifier Optimizer - Deep Learning Neural Network - Frequency Ratio (ICO-DLNN-FR) and Iterative Classifier Optimizer - Multilayer Perceptron - Frequency Ratio (ICO-MLP-FR). The first stage of the manuscript consisted of the collection and processing of the geodatabase needed in the present study. The geodatabase comprises a number of 14 flood predictors and 132 known flood locations. The Correlation-based Feature Selection (CFS) method was used in order to assess the prediction capacity of the 14 predictors in terms of flood susceptibility estimation. The training and validation of the three ensemble models constitute the next stage of the scientific workflow. Several statistical metrics and ROC curve method were involved in the evaluation of the model's performance and accuracy. According to ROC curves all the models achieved high performances since their AUC had values above 0.89. ICO-DLNN-FR proved to be the most accurate model (AUC = 0.959). The outcomes of the study can be used to guide future flood risk management and sustainable land-use planning in the designated area.

Predicting sustainable arsenic mitigation using machine learning techniques.

This study evaluates state-of-the-art machine learning models in predicting the most sustainable arsenic mitigation preference. A Gaussian distribution-based Naïve Bayes (NB) classifier scored the highest Area Under the Curve (AUC) of the Receiver Operating Characteristic curve (0.82), followed by Nu Support Vector Classification (0.80), and K-Neighbors (0.79). Ensemble classifiers scored higher than 70% AUC, with Random Forest being the top performer (0.77), and Decision Tree model ranked fourth with an AUC of 0.77. The multilayer perceptron model also achieved high performance (AUC=0.75). Most linear classifiers underperformed, with the Ridge classifier at the top (AUC=0.73) and perceptron at the bottom (AUC=0.57). A Bernoulli distribution-based Naïve Bayes classifier was the poorest model (AUC=0.50). The Gaussian NB was also the most robust ML model with the slightest variation of Kappa score on training (0.58) and test data (0.64). The results suggest that nonlinear or ensemble classifiers could more accurately understand the complex relationships of socio-environmental data and help develop accurate and robust prediction models of sustainable arsenic mitigation. Furthermore, Gaussian NB is the best option when data is scarce.

Investigation on factors affecting early strength of high-performance concrete by Gaussian Process Regression.

This study aims to investigate the influence of all the mixture components of high-performance concrete (HPC) on its early compressive strength, ranging from 1 to 14 days. To this purpose, a Gaussian Process Regression (GPR) algorithm was first constructed using a database gathered from the available literature. The database included the contents of cement, blast furnace slag (BFS), fly ash (FA), water, superplasticizer, coarse, fine aggregates, and testing age as input variables to predict the output of the problem, which was the early compressive strength. Several standard statistical criteria, such as the Pearson correlation coefficient, root mean square error and mean absolute error, were used to quantify the performance of the GPR model. To analyze the sensitivity and influence of the HPC mixture components, partial dependence plots analysis was conducted with both one-dimensional and two-dimensional. Firstly, the results showed that the GPR performed well in predicting the early strength of HPC. Second, it was determined that the cement content and testing age of HPC were the most sensitive and significant elements affecting the early strength of HPC, followed by the BFS, water, superplasticizer, FA, fine aggregate, and coarse aggregate contents. To put it simply, this research might assist engineers select the appropriate amount of mixture components in the HPC production process to obtain the necessary early compressive strength.

Cross-country connectedness in inflation and unemployment: measurement and macroeconomic consequences.

We bring the notion of connectedness (Diebold and Yilmaz, Int J Forecast 28(1):57-66 2012) to a set of two critical macroeconomic variables as inflation and unemployment. We focus on the G7 economies plus Spain, and use monthly data-high-frequency data in a macro setting-to explore the extent and consequences of total and directional volatility spillovers across variables and countries. We find that total connectedness is larger for prices (58.28%) than for unemployment (41.81%). We also identify asymmetries per country that result in higher short-run Phillips curve trade-offs in recessions and lower trade-offs in expansions. Besides, by exploring time-varying connectedness (resulting from country-specific shocks), we find that volatility spillovers magnify in periods of common economic turmoil such as the Global Financial Crisis. Our results call for an enhancement of international macroeconomic policy coordination. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s00181-021-02052-0.

On Random Subspace Optimization-Based Hybrid Computing Models Predicting the California Bearing Ratio of Soils.

The California Bearing Ratio (CBR) is an important index for evaluating the bearing capacity of pavement subgrade materials. In this research, random subspace optimization-based hybrid computing models were trained and developed for the prediction of the CBR of soil. Three models were developed, namely reduced error pruning trees (REPTs), random subsurface-based REPT (RSS-REPT), and RSS-based extra tree (RSS-ET). An experimental database was compiled from a total of 214 soil samples, which were classified according to AASHTO M 145, and included 26 samples of A-2-6 (clayey gravel and sand soil), 3 samples of A-4 (silty soil), 89 samples of A-6 (clayey soil), and 96 samples of A-7-6 (clayey soil). All CBR tests were performed in soaked conditions. The input parameters of the models included the particle size distribution, gravel content (G), coarse sand content (CS), fine sand content (FS), silt clay content (SC), organic content (O), liquid limit (LL), plastic limit (PL), plasticity index (PI), optimum moisture content (OMC), and maximum dry density (MDD). The accuracy of the developed models was assessed using numerous performance indexes, such as the coefficient of determination, relative error, MAE, and RMSE. The results show that the highest prediction accuracy was obtained using the RSS-based extra tree optimization technique.

Development of Artificial Neural Network for prediction of radon dispersion released from Sinquyen Mine, Vietnam.

Understanding the radon dispersion released from this mine are important targets as radon dispersion is used to assess radiological hazard to human. In this paper, the main objective is to develop and optimize a machine learning model namely Artificial Neural Network (ANN) for quick and accurate prediction of radon dispersion released from Sinquyen mine, Vietnam. For this purpose, a total of million data collected from the study area, which includes input variables (the gamma data of uranium concentration with 3 × 3m grid net survey inside mine, 21 of CR-39 detectors inside dwellings surrounding mine, and gamma dose at 1 m from ground surface data) and an output variable (radon dispersion) were used for training and validating the predictive model. Various validation methods namely coefficient of determination (R2), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) were used. In addition, Partial dependence plots (PDP) was used to evaluate the effect of each input variable on the predictive results of output variable. The results show that ANN performed well for prediction of radon dispersion, with low values of error (i.e., R2 = 0.9415, RMSE = 0.0589, and MAE = 0.0203 for the testing dataset). The increase of number of hidden layers in ANN structure leads the increase of accuracy of the predictive results. The sensitivity results show that all input variables govern the dispersion radon activity with different amplitudes and fitted with different equations but the gamma dose is the most influenced and important variable in comparison with strike, distance and uranium concentration variables for prediction of radon dispersion.

Groundwater Potential Mapping Using GIS-Based Hybrid Artificial Intelligence Methods.

Groundwater is one of the major valuable water resources for the use of communities, agriculture, and industries. In the present study, we have developed three novel hybrid artificial intelligence (AI) models which is a combination of modified RealAdaBoost (MRAB), bagging (BA), and rotation forest (RF) ensembles with functional tree (FT) base classifier for the groundwater potential mapping (GPM) in the basaltic terrain at DakLak province, Highland Centre, Vietnam. Based on the literature survey, these proposed hybrid AI models are new and have not been used in the GPM of an area. Geospatial techniques were used and geo-hydrological data of 130 groundwater wells and 12 topographical and geo-environmental factors were used in the model studies. One-R Attribute Evaluation feature selection method was used for the selection of relevant input parameters for the development of AI models. The performance of these models was evaluated using various statistical measures including area under the receiver operation curve (AUC). Results indicated that though all the hybrid models developed in this study enhanced the goodness-of-fit and prediction accuracy, but MRAB-FT (AUC = 0.742) model outperformed RF-FT (AUC = 0.736), BA-FT (AUC = 0.714), and single FT (AUC = 0.674) models. Therefore, the MRAB-FT model can be considered as a promising AI hybrid technique for the accurate GPM. Accurate mapping of the groundwater potential zones will help in adequately recharging the aquifer for optimum use of groundwater resources by maintaining the balance between consumption and exploitation.

Flash-Flood Potential Mapping Using Deep Learning, Alternating Decision Trees and Data Provided by Remote Sensing Sensors.

There is an evident increase in the importance that remote sensing sensors play in the monitoring and evaluation of natural hazards susceptibility and risk. The present study aims to assess the flash-flood potential values, in a small catchment from Romania, using information provided remote sensing sensors and Geographic Informational Systems (GIS) databases which were involved as input data into a number of four ensemble models. In a first phase, with the help of high-resolution satellite images from the Google Earth application, 481 points affected by torrential processes were acquired, another 481 points being randomly positioned in areas without torrential processes. Seventy percent of the dataset was kept as training data, while the other 30% was assigned to validating sample. Further, in order to train the machine learning models, information regarding the 10 flash-flood predictors was extracted in the training sample locations. Finally, the following four ensembles were used to calculate the Flash-Flood Potential Index across the Bâsca Chiojdului river basin: Deep Learning Neural Network-Frequency Ratio (DLNN-FR), Deep Learning Neural Network-Weights of Evidence (DLNN-WOE), Alternating Decision Trees-Frequency Ratio (ADT-FR) and Alternating Decision Trees-Weights of Evidence (ADT-WOE). The model's performances were assessed using several statistical metrics. Thus, in terms of Sensitivity, the highest value of 0.985 was achieved by the DLNN-FR model, meanwhile the lowest one (0.866) was assigned to ADT-FR ensemble. Moreover, the specificity analysis shows that the highest value (0.991) was attributed to DLNN-WOE algorithm, while the lowest value (0.892) was achieved by ADT-FR. During the training procedure, the models achieved overall accuracies between 0.878 (ADT-FR) and 0.985 (DLNN-WOE). K-index shows again that the most performant model was DLNN-WOE (0.97). The Flash-Flood Potential Index (FFPI) values revealed that the surfaces with high and very high flash-flood susceptibility cover between 46.57% (DLNN-FR) and 59.38% (ADT-FR) of the study zone. The use of the Receiver Operating Characteristic (ROC) curve for results validation highlights the fact that FFPIDLNN-WOE is characterized by the most precise results with an Area Under Curve of 0.96.

A Novel Hybrid Model Based on a Feedforward Neural Network and One Step Secant Algorithm for Prediction of Load-Bearing Capacity of Rectangular Concrete-Filled Steel Tube Columns.

In this study, a novel hybrid surrogate machine learning model based on a feedforward neural network (FNN) and one step secant algorithm (OSS) was developed to predict the load-bearing capacity of concrete-filled steel tube columns (CFST), whereas the OSS was used to optimize the weights and bias of the FNN for developing a hybrid model (FNN-OSS). For achieving this goal, an experimental database containing 422 instances was firstly gathered from the literature and used to develop the FNN-OSS algorithm. The input variables in the database contained the geometrical characteristics of CFST columns, and the mechanical properties of two CFST constituent materials, i.e., steel and concrete. Thereafter, the selection of the appropriate parameters of FNN-OSS was performed and evaluated by common statistical measurements, for instance, the coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MAE). In the next step, the prediction capability of the best FNN-OSS structure was evaluated in both global and local analyses, showing an excellent agreement between actual and predicted values of the load-bearing capacity. Finally, an in-depth investigation of the performance and limitations of FNN-OSS was conducted from a structural engineering point of view. The results confirmed the effectiveness of the FNN-OSS as a robust algorithm for the prediction of the CFST load-bearing capacity.

Cost-Effective Approaches Based on Machine Learning to Predict Dynamic Modulus of Warm Mix Asphalt with High Reclaimed Asphalt Pavement.

Warm mix asphalt (WMA) technology, taking advantage of reclaimed asphalt pavements, has gained increasing attention from the scientific community. The determination of technical specifications of such a type of asphalt concrete is crucial for pavement design, in which the asphalt concrete dynamic modulus (E*) of elasticity is amongst the most critical parameters. However, the latter could only be determined by complicated, costly, and time-consuming experiments. This paper presents an alternative cost-effective approach to determine the dynamic elastic modulus (E*) of WMA based on various machine learning-based algorithms, namely the artificial neural network (ANN), support vector machine (SVM), Gaussian process regression (GPR), and ensemble boosted trees (Boosted). For this, a total of 300 samples were fabricated by warm mix asphalt technology. The mixtures were prepared with 0%, 20%, 30%, 40%, and 50% content of reclaimed asphalt pavement (RAP) and modified bitumen binder using Sasobit and Zycotherm additives. The dynamic elastic modulus tests were conducted by varying the temperature from 10 °C to 50 °C at different frequencies from 0.1 Hz to 25 Hz. Various common quantitative indications, such as root mean square error (RMSE), mean absolute error (MAE), and correlation coefficient (R) were used to validate and compare the prediction capability of different models. The results showed that machine learning models could accurately predict the dynamic elastic modulus of WMA using up to 50% RAP and fabricated by warm mix asphalt technology. Out of these models, the Boosted algorithm (R = 0.9956) was found as the best predictor compared with those obtained by ANN-LMN (R = 0.9954), SVM (R = 0.9654), and GPR (R= 0.9865). Thus, it could be concluded that Boosted is a promising cost-effective tool for the prediction of the dynamic elastic modulus (E*) of WMA. This study might help in reducing the cost of laboratory experiments for the determination of the dynamic modulus (E*).

Bifunctional and binder-free S-doped Ni-P nanospheres electrocatalyst fabricated by pulse electrochemical deposition method for overall water splitting.

Due to their excellent electrocatalytic properties, transition metal phosphides have been considered as desirable and cost-effective electrocatalysts in recent years. However, in many cases, the synthesis of phosphide-based nanostructures requires expensive conditions and toxic phosphorous-containing compounds. Therefore, the emergence of an economical and eco-friendly method for creating phosphides-based nanostructures can be very effective. Here, S-doped Ni-P nanospheres were created using a novel pulse electrochemical deposition method, and its hydrogen evolution reaction (HER), as well as oxygen evolution reaction (OER) electrocatalytic activity and stability were investigated. Inspired by the high electrochemically active surface area, the synergistic effect between S and P, the rapid detachment of the gasses from the surface and thus the lowered resistance resulted from bubbles pinning, and ultimately the increased wettability due to nanostructuring, the electrode exhibited outstanding electrocatalytic activity for HER and OER processes. The electrode requires only 55 mV and 229 mV overproduction to be able to afford the current density of 10 mA.cm-2 for HER and OER processes, respectively. Furthermore, at a current density of 100 mA.cm-2, these electrodes showed minor changes in potential, indicating the superb electrocatalytic stability of the synthesized electrode. Moreover, in the overall water splitting process, the S-doped Ni-P electrode requires only 1.51 V to generate a current of 10 mA.cm-2. The results of this study indicate the successful use of pulse electrochemical deposition method to create active electrocatalysts.

Parametric Investigation of Particle Swarm Optimization to Improve the Performance of the Adaptive Neuro-Fuzzy Inference System in Determining the Buckling Capacity of Circular Opening Steel Beams.

In this paper, the main objectives are to investigate and select the most suitable parameters used in particle swarm optimization (PSO), namely the number of rules (nrule), population size (npop), initial weight (wini), personal learning coefficient (c1), global learning coefficient (c2), and velocity limits (fv), in order to improve the performance of the adaptive neuro-fuzzy inference system in determining the buckling capacity of circular opening steel beams. This is an important mechanical property in terms of the safety of structures under subjected loads. An available database of 3645 data samples was used for generation of training (70%) and testing (30%) datasets. Monte Carlo simulations, which are natural variability generators, were used in the training phase of the algorithm. Various statistical measurements, such as root mean square error (RMSE), mean absolute error (MAE), Willmott's index of agreement (IA), and Pearson's coefficient of correlation (R), were used to evaluate the performance of the models. The results of the study show that the performance of ANFIS optimized by PSO (ANFIS-PSO) is suitable for determining the buckling capacity of circular opening steel beams, but is very sensitive under different PSO investigation and selection parameters. The findings of this study show that nrule = 10, npop = 50, wini = 0.1 to 0.4, c1 = [1, 1.4], c2 = [1.8, 2], fv = 0.1, which are the most suitable selection values to ensure the best performance for ANFIS-PSO. In short, this study might help in selection of suitable PSO parameters for optimization of the ANFIS model.

Particulate matter concentration from open-cut coal mines: A hybrid machine learning estimation.

Particulate matter (PM) emission is one of the leading environmental pollution issues associated with the coal mining industry. Before any control techniques can be employed, however, an accurate prediction of PM concentration is desired. Towards this end, this work aimed to provide an accurate estimation of PM concentration using a hybrid machine-learning technique. The proposed predictive model was based on the hybridazation of random forest (RF) model particle swarm optimization (PSO) for estimating PM concentration. The main objective of hybridazing the PSO was to tune the hyper-parameters of the RF model. The hybrid method was applied to PM data collected from an open-cut coal mine in northern China, the Haerwusu Coal Mine. The inputs selected were wind direction, wind speed, temperature, humidity, noise level and PM concentration at 5 min before. The outputs selected were the current concentration of PM2.5 (particles with an aerodynamic diameter smaller than 2.5 µm), PM10 (particles with an aerodynamic diameter smaller than 10 µm) and total suspended particulate (TSP). A detailed procedure for the implementation of the RF_PSO was presented and the predictive performance was analyzed. The results show that the RF_PSO could estimate PM concentration with a high degree of accuracy. The Pearson correlation coefficients among the average estimated and measured PM data were 0.91, 0.84 and 0.86 for the PM2.5, PM10 and TSP datasets, respectively. The relative importance analysis shows that the current PM concentration was mainly influenced by PM concentration at 5 min before, followed by humidity > temperature ≈ noise level > wind speed > wind direction. This study presents an efficient and accurate way to estimate PM concentration, which is fundamental to the assessment of the atmospheric quality risks emanating from open-cut mining and the design of dust removal techniques.

Groundwater Potential Mapping Combining Artificial Neural Network and Real AdaBoost Ensemble Technique: The DakNong Province Case-study, Vietnam.

: The main aim of this study is to assess groundwater potential of the DakNong province, Vietnam, using an advanced ensemble machine learning model (RABANN) that integrates Artificial Neural Networks (ANN) with RealAdaBoost (RAB) ensemble technique. For this study, twelve conditioning factors and wells yield data was used to create the training and testing datasets for the development and validation of the ensemble RABANN model. Area Under the Receiver Operating Characteristic (ROC) curve (AUC) and several statistical performance measures were used to validate and compare performance of the ensemble RABANN model with the single ANN model. Results of the model studies showed that both models performed well in the training phase of assessing groundwater potential (AUC ≥ 0.7), whereas the ensemble model (AUC = 0.776) outperformed the single ANN model (AUC = 0.699) in the validation phase. This demonstrated that the RAB ensemble technique was successful in improving the performance of the single ANN model. By making minor adjustment in the input data, the ensemble developed model can be adapted for groundwater potential mapping of other regions and countries toward more efficient water resource management. The present study would be helpful in improving the groundwater condition of the area thus in solving water borne disease related health problem of the population.

Different sampling strategies for predicting landslide susceptibilities are deemed less consequential with deep learning.

Predictive capability of landslide susceptibilities is assumed to be varied with different sampling techniques, such as (a) the landslide scarp centroid, (b) centroid of landslide body, (c) samples of the scrap region representing the scarp polygon, and (d) samples of the landslide body representing the entire landslide body. However, new advancements in statistical and machine learning algorithms continuously being updated the landslide susceptibility paradigm. This paper explores the predictive performance power of different sampling techniques in landslide susceptibility mapping in the wake of increased usage of artificial intelligence. We used logistic regression (LR), neural network (NNET), and deep learning neural network (DNN) model for testing and validation of the models. The tests were applied to the 2018 Hokkaido Earthquake affected areas using a set of 11 predictor variables (seismic, topographic, and hydrological). We found that the prediction rates are inconsequential with the DNN model irrespective of the sampling technique (AUC: 0.904 - 0.919). Whereas, testing with LR (AUC: 0.825 - 0.785) and NNET (AUC: 0.882 - 0.858) produces larger differences in the accuracies between the four datasets. Nonetheless, the highest success rates were obtained for samples within the landslide scarp area. The analogy was then validated with a published landslide inventory from the 2015 Gorkha earthquake. We, therefore, suggest that DNN models as an appropriate technique to increase the predictive performance of landslide susceptibilities if the landslide scarp and body are not characterized properly in an inventory.

Investigation and Optimization of the C-ANN Structure in Predicting the Compressive Strength of Foamed Concrete.

Development of Foamed Concrete (FC) and incessant increases in fabrication technology have paved the way for many promising civil engineering applications. Nevertheless, the design of FC requires a large number of experiments to determine the appropriate Compressive Strength (CS). Employment of machine learning algorithms to take advantage of the existing experiments database has been attempted, but model performance can still be improved. In this study, the performance of an Artificial Neural Network (ANN) was fully analyzed to predict the 28 days CS of FC. Monte Carlo simulations (MCS) were used to statistically analyze the convergence of the modeled results under the effect of random sampling strategies and the network structures selected. Various statistical measures such as Coefficient of Determination (R2), Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE) were used for validation of model performance. The results show that ANN is a highly efficient predictor of the CS of FC, achieving a maximum R2 value of 0.976 on the training part and an R2 of 0.972 on the testing part, using the optimized C-ANN-[3,4,5,1] structure, which compares with previous published studies. In addition, a sensitivity analysis using Partial Dependence Plots (PDP) over 1000 MCS was also performed to interpret the relationship between the input parameters and 28 days CS of FC. Dry density was found as the variable with the highest impact to predict the CS of FC. The results presented could facilitate and enhance the use of C-ANN in other civil engineering-related problems.