Bagging-based positive-unlabeled learning algorithm with Bayesian hyperparameter optimization for three-dimensional mineral potential mapping

Zhang, Zhiqiang; Wang, Gongwen; Liu, Chong; Cheng, Lingpeng; Sha, Deming

doi:10.1016/j.cageo.2021.104817

Cited by 27 publications

(24 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When a random forest initially has only one base learner, it is a decision tree model with random selection generating perturbations, which is characterized by poor performance, poor fitting, low generalization ability, etc. Multiple stochastic decision tree models are integrated in parallel by Bagging, 31 and the results are predicted by voting or averaging between each base learning model. In this study, random forest parameters were determined by grid search, the number of base models was 100, the quality of attribute partitioning was measured using mean square error (MSE), and the default values were used for the remaining parameters.…”

Section: Data Processing and Machine Learning Modelsmentioning

confidence: 99%

Application of Machine Learning in a Mineral Leaching Process─Taking Pyrolusite Leaching as an Example

Zhang

et al. 2022

ACS Omega

View full text Add to dashboard Cite

In this study, several machine learning models were used to analyze the process variables of electric-field-enhanced pyrolusite leaching and predict the leaching rate of manganese, and the applicability of those models in the leaching process of hydrometallurgy was compared. It showed that there was no correlation between the six leaching conditions; in addition to the leaching time, the concentrations of sulfuric acid and ferrous sulfate had great influences on the leaching of pyrolusite. The results of the prediction models showed that the support vector regression model has the best prediction performance, with regression index (R 2) = 0.92 and mean square error = 25.04, followed by the gradient boosting regression model (R 2 > 0.85). In this research, machine learning models were applied to the optimization of the manganese leaching process, and the research process and methods were also applicable to other hydrometallurgical processes for majorization and result prediction.

show abstract

Section: Data Processing and Machine Learning Modelsmentioning

confidence: 99%

Application of Machine Learning in a Mineral Leaching Process─Taking Pyrolusite Leaching as an Example

Zhang

et al. 2022

ACS Omega

View full text Add to dashboard Cite

show abstract

“…However, the uncertainty of geological conditions and field measurement errors significantly impact the inversion results. The inversion of hydraulic conductivity can be transformed into an optimization problem by designing an appropriate objective function with a mathematical seepage model [ 12 , 13 ]. Deep learning methods can effectively improve computational efficiency and accuracy of optimization problems.…”

Section: Introductionmentioning

confidence: 99%

A New Method for Inversion of Dam Foundation Hydraulic Conductivity Using an Improved Genetic Algorithm Coupled with an Unsaturated Equivalent Continuum Model and Its Application

Peng

Shen

et al. 2023

Materials

View full text Add to dashboard Cite

Seepage is a main cause of dam failure, and its stability analysis is the focus of a dam’s design, construction, and management. Because a geological survey can only determine the range of a dam foundation’s hydraulic conductivity, hydraulic conductivity inversion is crucial in engineering. However, current inversion methods of dam hydraulic conductivity are either not accurate enough or too complex to be directly used in engineering. Therefore, this paper proposes a new method for the inversion of hydraulic conductivity with high application value in hydraulic engineering using an improved genetic algorithm coupled with an unsaturated equivalent continuum model (IGA–UECM). This method is implemented by a new code that fully considers engineering applicability. In addition to overcoming the premature convergence shortcomings of traditional genetic algorithms, it converges faster than Bayesian optimization and tree-structured Parzen estimator inversion algorithms. This method is verified by comparing the water head from drilling exploration and inversion. The results of the inversion are used to study the influence of a cement grouting curtain layout scheme on the seepage field of the Hami concrete-face rockfill dam in China, which is used as an engineering application case of the IGA–UECM. The law of the seepage field is reasonable, which verifies the validity of the IGA–UECM. The new inversion method of hydraulic conductivity and the proposed cement grouting curtain layout in this study offer possible strategies for the design, construction, and management of concrete-face rockfill dams.

show abstract

“…With the development of computational technology and the continuous accumulation of geosciences datasets, mineral exploration has developed from near-surface to subsurface, from 2D to 3D, from qualitative to quantitative (Yuan et al, 2019;Zhang Z Q et al, 2021). 3D mineral prospectivity modeling is developed based on the 3D geological modeling and they are both widely applied in the mineral exploration (Houlding, 1994;Li et al, 2015;Xiao et al, 2015;Li et al, 2016;Wang G W et al, 2017;Yang et al, 2017;Mao et al, 2019;Wang et al, 2021;Zhang Z Q et al, 2021;Gao et al, 2023). Since the 1990s, various knowledge-and data-driven learning models have been applied to conduct mineral prospectivity modeling.…”

Section: Introductionmentioning

confidence: 99%

“…Based on the supervised algorithms, some derivative algorithms such as the semi-supervised random forests, one-class support vector machine and isolation forest have occurred (Chen and Wu, 2017;Chen and Wu, 2019;Wang et al, 2020). RF and XGBoost (Chen and Guestrin, 2016) with excellent performance were used as the base learners of baggingbased positive-unlabeled learning algorithm (Zhang Z Q et al, 2021;Gao et al, 2023). In addition, deep learning is outstanding in the field of 2D mineral exploration (Zuo et al, 2019;Yang et al, 2023) and has been applied to 3D mineral exploration (Li et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

3D mineral exploration Cu-Zn targeting with multi-source geoscience datasets in the Weilasituo-bairendaba district, Inner Mongolia, China

Gao

Wang

et al. 2023

Front. Earth Sci.

Self Cite

View full text Add to dashboard Cite

The Weilasituo-bairendaba district is located at the eastern end of the Central Asian Orogenic Belt, which is an important component of the Cu-Pb-Zn polymetallic metallogenic belt on the Western slope of the Greater Xing’an Range in Inner Mongolia, China. The known Cu-Zn deposits such as the Weilasituo Cu-Zn deposit and the Bairendaba Ag-Pb-Zn deposit are the same tectonic-magmatic product. The district’s structure framework consists of the NE-trending regional faults, while the secondary faults provide channels and space for mineralization. The ore-bearing rocks are either Baoyintu Group gneisses or quartz diorites. The typical Cu-Zn deposits exhibit obvious Cu, Pb, Zn geochemical anomaly as well as obvious magnetic anomaly. The district-scale two-dimensional (2D) mineral prospectivity modeling has been reported. Nowadays, three-dimensional (3D) mineral prospectivity modeling is necessary and urgent. Integrated deposit geology and accumulated exploration data, the above four exploration criteria (regional fault, secondary fault, geochemical anomaly and magnetic susceptibility) are used for 3D mineral prospectivity modeling. Filtering (upward continuation, low pass filtering, two-dimensional empirical mode decomposition), magnetic inversion and 3D modeling techniques were used to construct geological models. Excellent machine learning algorithms such as random forest (RF) and XGBoost are applied. The two machine learning methods confirm each other to improve the accuracy of 3D mineral prospectivity modeling. In this paper, repeated random sampling and Bayesian optimization are combined to construct and tune models. This joint method can avoid the contingency caused by random sampling of negative samples, and can also realize automatic optimization of hyperparameters. The optimal models (RF28 and XGBoost11) were selected among thirty repeated training models for mineral prospectivity modeling. The obtained areas under the ROC curves of RF28 and XGBoost11 were 0.987 and 0.986, respectively. The prediction-area (P-A) plot and C-A fractal were used to delineate targets and grade targets. The targets were divided into Ⅰ-level targets and Ⅱ-level targets. The I- and II-targets are not only highly consistent with the known Cu-Zn deposits, but also exhibit obvious ore-forming geological features. The 3D targets are beneficial for Cu-Zn exploration in the Weilasituo-bairendaba district.

show abstract

Bagging-based positive-unlabeled learning algorithm with Bayesian hyperparameter optimization for three-dimensional mineral potential mapping

Cited by 27 publications

References 48 publications

Application of Machine Learning in a Mineral Leaching Process─Taking Pyrolusite Leaching as an Example

Application of Machine Learning in a Mineral Leaching Process─Taking Pyrolusite Leaching as an Example

A New Method for Inversion of Dam Foundation Hydraulic Conductivity Using an Improved Genetic Algorithm Coupled with an Unsaturated Equivalent Continuum Model and Its Application

3D mineral exploration Cu-Zn targeting with multi-source geoscience datasets in the Weilasituo-bairendaba district, Inner Mongolia, China

Contact Info

Product

Resources

About