A Physically Constrained Variational Autoencoder for Geochemical Pattern Recognition

Xiong, Yihui; Luo, Zijing; Wang, Xueqiu

doi:10.1007/s11004-021-09979-1

Cited by 37 publications

(4 citation statements)

References 90 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Conversely, Caledonian granite, Yanshanian granite, and fault distribution had little or no influence on the prediction result in the RF model. The learning results of the five models for these variables are consistent with previous research that shows the anomalies of W, Sn, Cd, Bi, and Be are closely related to tungsten polymetallic deposits in the Gannan region (Liu et al., 2014, 2019; Xiong et al., 2021). Additionally, these elements, especially W and Sn, have similar geochemical properties and are closely related to the residual magma, resulting from partial melting of the continental crust (Liu et al., 2014).…”

Section: Resultssupporting

confidence: 89%

Mineral Prospectivity Mapping of Tungsten Polymetallic Deposits Using Machine Learning Algorithms and Comparison of Their Performance in the Gannan Region, China

Lou,

Liu

2023

Earth and Space Science

View full text Add to dashboard Cite

The current study aims at assessing the capabilities of five machine learning models in terms of mapping tungsten polymetallic prospectivity in the Gannan region, China. The five models include logistic regression (LR), support vector machine (SVM), random forest, convolutional neural network (CNN), and light gradient boosting machine (LGBM) models. Three types of data sets including geochemical data, lithostratigraphic contacts, and faults were employed to generate 16 evidential maps that were used to build the machine learning models. Tungsten polymetallic deposits were randomly separated into two parts: 80% for training and 20% for validating. Performance of the models was evaluated through receiver operating characteristic and K‐fold cross validation, with an emphasis on the variable influence within different machine learning methods. The results show that the models are especially sensitive to the chemical elements: Be, Bi, Pb, and Cd, implying that they are closely related to tungsten polymetallic mineralization. Compared to other models, the LGBM and CNN models performed best, while the LR model was the most stable. The results also indicated that the CNN model can predict the maximum number of tungsten polymetallic deposits within the minimum area based on the prediction‐area plot analysis, while the SVM model can capture the least amount of tungsten polymetallic deposits within the largest area. Finally, 18 prospective areas were delineated according to the prediction results of the machine learning models, which provided important guidance for tungsten polymetallic exploration in the study area.

show abstract

Section: Resultssupporting

confidence: 89%

Mineral Prospectivity Mapping of Tungsten Polymetallic Deposits Using Machine Learning Algorithms and Comparison of Their Performance in the Gannan Region, China

Lou,

Liu

2023

Earth and Space Science

View full text Add to dashboard Cite

show abstract

“…Conversely, Caledonian granite, Yanshanian granite, and fault distribution had little or no influence in the RF model. The learning results of the five models for these variables are consistent with the previous research conclusion that the anomalies of W, Sn, Cd, Bi, Be, Pb, and other elements are closely related to the tungsten polymetallic deposits in southern Jiangxi (Liu et al, 2014a(Liu et al, , 2019Xiong et al, 2021). Additionally, these elements, specifically W and Sn, have similar geochemical properties and closely relate to the residual magma, resulting from partial melting of the continental crust (Liu et al, 2014a).…”

Section: Variables Contribution Analysissupporting

confidence: 89%

“…Genetically, most of the Nanling region W-Sn-polymetallic deposits have a relationship with Mesozoic granitoids and can be divided into 10 types (Wang et al, 2020b). Using a physically constrained variational automatic encoder (VAE), Xiong et al (2021) identified geochemical patterns associated with tungsten polymetallic mineralization in the Nanling Mountain Range.…”

Section: Granite Massmentioning

confidence: 99%

Mineral prospectivity mapping of tungsten polymetallic deposits using machine learning algorithms and comparison of their performance in the Gannan region, China

Liu

Yonghang

2022

Preprint

View full text Add to dashboard Cite

The current study aimed at assessing the capabilities of five machine learning models in term of mapping tungsten polymetallic prospectivity in the Gannan region of China. The five models include logistic regression (LR), support vector machine (SVM), random forest (RF), convolutional neural network (CNN), and light gradient boosting machine (LGBM) models. Geochemical, lithostratigraphic, and structural datasets were used to generate 16 evidential maps, which were integrated into the machine learning models. Tungsten polymetallic deposits were randomly separated into two parts: 80% for training and 20% for validating. Performances of the models were evaluated through receiver operating characteristic (ROC) and K-fold cross validation, with an emphasis on the variable influence within different machine learning methods. The results show that the models are especially sensitive to the chemical elements: Be, Bi, Pb and Cd, implying that these are closely related to tungsten polymetallic mineralization. Compared to other models, the LGBM and CNN models performed best, while the LR model was the most stable. The results also indicated that the CNN model can predict maximum known deposits within a minimum area, based on the prediction-area plot analysis of the five models, while the RF model can capture the most well-known deposits within the smallest study area. Finally, eighteen prospective areas were delineated according to the predicting results of the machine learning models, which will provide important guidance for further tungsten polymetallic exploration and associated studies.

show abstract

“…The idea of Deep Auto-encoder network for anomaly detection in geosciences is to learn the reconstruction ability of anomaly-free samples, and to achieve anomaly detection by reconstructing anomaly samples with a large error. Related achievements include the integration of auto-encoder network with density-based spatial clustering for geochemical anomaly detection [52], physically constrained variational auto-encoder for geochemical pattern recognition [53], and detection of geochemical anomalies related to mineralization using the GANomaly network [54]. DAE is an unsupervised learning method, which achieves more efficient detection of ore-induced comprehensive anomalies, and does not need to manually label positive and negative samples, saving valuable labor costs.…”

Section: Introductionmentioning

confidence: 99%

3D Mineral Prospectivity Mapping of Zaozigou Gold Deposit, West Qinling, China: Deep Learning-Based Mineral Prediction

Liu

Xie

et al. 2022

Minerals

View full text Add to dashboard Cite

This paper focuses on the scientific problem of quantitative mineralization prediction at large depth in the Zaozigou gold deposit, west Qinling, China. Five geological and geochemical indicators are used to establish geological and geochemical quantitative prediction model. Machine learning and Deep learning algorithms are employed for 3D Mineral Prospectivity Mapping (MPM). Especially, the Student Teacher Ore-induced Anomaly Detection (STOAD) model is proposed based on the knowledge distillation (KD) idea combined with Deep Auto-encoder (DAE) network model. Compared to DAE, STOAD uses three outputs for anomaly detection and can make full use of information from multiple levels of data for greater overall robustness. The results show that the quantitative mineral resources prediction by applying the STOAD model has a good performance, where the value of Area Under Curve (AUC) is 0.97. Finally, three main mineral exploration targets are delineated for further investigation.

show abstract

A Physically Constrained Variational Autoencoder for Geochemical Pattern Recognition

Cited by 37 publications

References 90 publications

Mineral Prospectivity Mapping of Tungsten Polymetallic Deposits Using Machine Learning Algorithms and Comparison of Their Performance in the Gannan Region, China

Mineral Prospectivity Mapping of Tungsten Polymetallic Deposits Using Machine Learning Algorithms and Comparison of Their Performance in the Gannan Region, China

Mineral prospectivity mapping of tungsten polymetallic deposits using machine learning algorithms and comparison of their performance in the Gannan region, China

3D Mineral Prospectivity Mapping of Zaozigou Gold Deposit, West Qinling, China: Deep Learning-Based Mineral Prediction

Contact Info

Product

Resources

About