Abstract Growing an ensemble of decision trees and allowing them to vote for the most popular class produced a significant increase in classification accuracy land cover classification. The objective this study is present results obtained with random forest classifier compare its performance support vector machines (SVMs) terms accuracy, training time user defined parameters. Landsat Enhanced Thematic Mapper Plus (ETM+) data area UK seven different covers were used. Results from suggest that...

Abstract Support vector machines (SVM) represent a promising development in machine learning research that is not widely used within the remote sensing community. This paper reports results of two experiments which multi‐class SVMs are compared with maximum likelihood (ML) and artificial neural network (ANN) methods terms classification accuracy. The land cover use multispectral (Landsat‐7 ETM+) hyperspectral (DAIS) data, respectively, for test areas eastern England central Spain. Our show...

Support vector machines (SVM) are attractive for the classification of remotely sensed data with some claims that method is insensitive to dimensionality and, therefore, does not require a dimensionality-reduction analysis in preprocessing. Here, series analyses two hyperspectral sensor sets reveals accuracy by an SVM vary as function number features used. Critically, it shown may decline significantly (at 0.05 level statistical significance) addition features, particularly if small training...

Abstract This letter evaluates the effectiveness of a new kernel-based extreme learning machine (ELM) algorithm for land cover classification using both multi- and hyperspectral remote-sensing data. The results are compared with most widely used algorithms – support vector machines (SVMs). in terms ease use (in number user-defined parameters), accuracy computation cost. A radial basis kernel function was SVM extreme-learning to ensure compatibility comparison two algorithms. suggest that is...

This paper investigates the potential of support vector machines (SVM)-based classification approach to assess liquefaction from actual standard penetration test (SPT) and cone (CPT) field data. SVMs are based on statistical learning theory found work well in comparison neural networks several other applications. Both CPT SPT data sets is used with for predicting occurrence non-occurrence different input parameter combination. With sets, highest accuracy 96 97%, respectively, was achieved...

Abstract This paper investigates the potential of M5 model tree based regression approach to daily reference evapotranspiration using climatic data Davis station maintained by California irrigation Management Information System (CIMIS). Four inputs including solar radiation, average air temperature, relative humidity, and wind speed whereas calculated a relation provided CIMIS was used as output. To compare performance in predicting evapotranspiration, FAO–56 Penman–Monteith equation...

The accuracy of a conventional supervised classification is in part function the training set used, notably impacted by quantity and quality cases. Since it can be costly to acquire large number high cases, recent research has focused on methods that allow accurate from small sets. Previous work shown potential support vector machine (SVM) based classifiers. Here, relevance (RVM) sparse multinominal logistic regression (SMLR) approaches evaluated relative SVM classification. With both...

Abstract The present paper deals with performance evaluation of application three machine learning algorithms such as Deep neural network (DNN), Gradient boosting (GBM) and Extreme gradient (XGBoost) to evaluate the ground water indices over a study area Haryana state (India). To investigate applicability these models, two quality indices, namely Entropy Water Quality Index (EWQI) (WQI) are employed in study. Analysis results demonstrated that DNN has exhibited comparatively lower error...

Classification accuracy depends on a number of factors, which the nature training samples, bands used, classes to be identified relative spatial resolution image and properties classifier are most important. This paper evaluates effects these factors classification using test area in La Mancha, Spain. High spectral DAIS data were used compare performance four procedures (maximum likelihood, neural network, support vector machines decision tree). There was no evidence view that inevitably...

This letter presents the results of two different ensemble approaches to increase accuracy land cover classification using support vector machines. Finite approaches, based on boosting and bagging infinite created by embedding hypothesis in kernel machines, are discussed. Results suggest that approach provides a significant comparison radial basis function kernel‐based While finite works well comparable performance approach, whereas decreases Comparison terms computational cost suggests...

The accuracy of a map is dependent on the reference dataset used in its construction. Classification analyses thematic mapping can, for example, be sensitive to range sampling and data quality concerns. With particular focus latter, effects land cover classifications from airborne mapper are explored. Variations intensity effort highlighted that widely modelling studies; these may need accounting analyses. labelling was also key variable influencing accuracy. Accuracy varied with amount...

This study reports results of the Extreme Gradient Boosting (XGBoost) algorithm in comparison to Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN) generate cloud mask comprising six classes: low cloud, high shadow, ground, snow, water. Here, Baetens-Hagolle dataset WHUS2-CD was considered. Various texture features derived using GLCM, morphological profile analysis, bilateral filtering, deep Residual (Resnet) were used combination with spectral...

Abstract The present study explores the suitability of groundwater for drinking purpose and evaluates non-carcinogenic health risks children, women, men. For this purpose, 47 samples were collected analyzed physicochemical parameters, including nitrate concentration. results revealed that concentration varied from 15 to 85 mg/L 48.93% exceeded Bureau Indian Standards’ limits 45 mg/L. spatial map pollution index specifies most area lies in moderate high zones. Principal component analysis was...

Abstract This paper present the results of a support vector machine (SVM) technique and genetic algorithm (GA) using generalization error bounds derived for SVMs as fitness functions (SVM/GA) feature selection hyperspectral data. Results obtained with SVM/GA‐based were compared those produced by random forest‐ SVM‐based techniques in terms classification accuracy computational cost. The was 91.89%. number features selected 15. For comparison, use full set 65 91.76%. level achieved SVM/GA...

This note investigates the potential of support vector machines based regression approach to model static pile capacity from dynamic stress-wave data. A data set 105 prestressed precast high strength concrete spun pipe piles is used. Radial basis function and polynomial kernel were used total results compared with a generalized neural network approach. 81 train, whereas remaining 24 sets test created model. correlation coefficient value 0.977 was achieved by in comparison values 0.967 0.964...