Support vector machines and gradient boosting for graphical estimation of a slate deposit (original) (raw)
Related papers
Comparison of kriging and neural networks with application to the exploitation of a slate mine
Mathematical Geology, 2004
To carry out an efficient and effective exploitation of a slate mine, it is necessary to have detailed information about the production potential of the site. To assist us in estimating the quality of slate from a small set of drilling data within an unexploited portion of the mine, the following estimation techniques were applied: kriging, regularization networks (RN), multilayer perceptron (MLP) networks, and radial basis function (RBF) networks. Our numerical results for the test holes show that the best results were obtained using an RN (kriging) which takes into account the known anisotropy. Differing deposit configurations were obtained, depending on the method applied. Variations in the form of pockets were obtained when using a radial pattern with RBF, RN, and kriging models while a stratified pattern was obtained with the MLP model. Pockets are more suitable for a slate mine, which indicates that the selection of a technique should take account of the specific configuration of the deposit according to mineral type.
Support vector machine for multi-classification of mineral prospectivity areas
Computers & Geosciences, 2012
In this paper on mineral prospectivity mapping, a supervised classification method called Support Vector Machine (SVM) is used to explore porphyry-Cu deposits. Different data layers of geological, geophysical and geochemical themes are integrated to evaluate the Now Chun porphyry-Cu deposit, located in the Kerman province of Iran, and to prepare a prospectivity map for mineral exploration. The SVM method, a data-driven approach to pattern recognition, had a correct-classification rate of 52.38% for twenty-one boreholes divided into five classes. The results of the study indicated the capability of SVM as a supervised learning algorithm tool for the predictive mapping of mineral prospects. Multiclassification of the prospect for detailed study could increase the resolution of the prospectivity map and decrease the drilling risk.
Support vector machine: A tool for mapping mineral prospectivity
Computers & Geosciences, 2011
In this contribution, we describe an application of support vector machine (SVM), a supervised learning algorithm, to mineral prospectivity mapping. The free R package e1071 is used to construct a SVM with sigmoid kernel function to map prospectivity for Au deposits in western Meguma Terrain of Nova Scotia (Canada). The SVM classification accuracies of 'deposit' are 100%, and the SVM classification accuracies of the 'non-deposit' are greater than 85%. The SVM classifications of mineral prospectivity have 5-9% lower total errors, 13-14% higher false-positive errors and 25-30% lower false-negative errors compared to those of the WofE prediction. The prospective target areas predicted by both SVM and WofE reflect, nonetheless, controls of Au deposit occurrence in the study area by NE-SW trending anticlines and contact zones between Goldenville and Halifax Formations. The results of the study indicate the usefulness of SVM as a tool for predictive mapping of mineral prospectivity.
2005 International Conference on Machine Learning and Cybernetics, 2005
In the mining industry, effective use of geographic information systems (GIS) to identify new geographic locations that are favorable for mineral exploration is very important. However, definitive prediction of such location is not an easy task. In this paper, four different neural networks, namely, the Polynomial Neural Network (PNN), General Regression Neural Network (GRNN), Probabilistic Neural Network (PrNN) and Back Propagation Neural Network (BPNN) have been used to classify data corresponding to cells in a map grid into deposit cells and barren cells. These approaches were tested on the GIS mineral exploration data from the Kalgoorlie region of Western Australia. The performance of individual neural networks is compared based on simulation results. The results demonstrate various degrees of success for the networks and suggestions on how to integrate the results are discussed.
Journal of Intelligent Learning Systems and Applications, 2010
Traditional geostatistical estimation techniques have been used predominantly by the mining industry for ore reserve estimation. Determination of mineral reserve has posed considerable challenge to mining engineers due to the geological complexities of ore body formation. Extensive research over the years has resulted in the development of several state-of-the-art methods for predictive spatial mapping, which could be used for ore reserve estimation; and recent advances in the use of machine learning algorithms (MLA) have provided a new approach for solving the problem of ore reserve estimation. The focus of the present study was on the use of two MLA for estimating ore reserve: namely, neural networks (NN) and support vector machines (SVM). Application of MLA and the various issues involved with using them for reserve estimation have been elaborated with the help of a complex drill-hole dataset that exhibits the typical properties of sparseness and impreciseness that might be associated with a mining dataset. To investigate the accuracy and applicability of MLA for ore reserve estimation, the generalization ability of NN and SVM was compared with the geostatistical ordinary kriging (OK) method.
Geotechnical and Geological Engineering, 2008
Geospatial technology is increasing in demand for many applications in geosciences. Spatial variability of the bed/hard rock is vital for many applications in geotechnical and earthquake engineering problems such as design of deep foundations, site amplification, ground response studies, liquefaction, microzonation etc. In this paper, reduced level of rock at Bangalore, India is arrived from the 652 boreholes data in the area covering 220 km 2 . In the context of prediction of reduced level of rock in the subsurface of Bangalore and to study the spatial variability of the rock depth, Geostatistical model based on Ordinary Kriging technique, Artificial Neural Network (ANN) and Support Vector Machine (SVM) models have been developed. In Ordinary Kriging, the knowledge of the semi-variogram of the reduced level of rock from 652 points in Bangalore is used to predict the reduced level of rock at any point in the subsurface of the Bangalore, where field measurements are not available. A new type of crossvalidation analysis developed proves the robustness of the Ordinary Kriging model. ANN model based on multi layer perceptrons (MLPs) that are trained with Levenberg-Marquardt backpropagation algorithm has been adopted to train the model with 90% of the data available. The SVM is a novel type of learning machine based on statistical learning theory, uses regression technique by introducing loss function has been used to predict the reduced level of rock from a large set of data. In this study, a comparative study of three numerical models to predict reduced level of rock has been presented and discussed.
Prediction of Oil and Gas Reservoir Properties using Support Vector Machines
International Petroleum Technology Conference, 2011
Artificial Intelligence techniques have been used in petroleum engineering to predict various reservoir properties such as porosity, permeability, water saturation, lithofacie and wellbore stability. The most extensively used of these techniques is Artificial Neural Networks (ANN). More recent techniques such as Support Vector Machines (SVM) have featured in the literature with better performance indices. However, SVM has not been widely embraced in petroleum engineering as a possibly better alternative to ANN. ANN has been reported to have a lot of limitations such as its lack of global optima. On the other hand, SVM has been introduced as a generalization of the Tikhonov Regularization procedure that ensures its global optima and offers ease of training. This paper presents a comparative study of the application of ANN and SVM models in the prediction of porosity and permeability of oil and gas reservoirs with carbonate platforms. Six datasets obtained from oil and gas reservoirs in two different geographical locations were used for the training, testing and validation of the models using the stratified sampling approach rather than the conventional static method of data division. The results showed that the SVM model performed better than the popularly used Feed forward Back propagation ANN with higher correlation coefficients and lower root mean squared errors. The SVM was also faster in terms of execution time. Hence, this work presents SVM as a possible alternative to ANN, especially, in the characterization of oil and gas reservoir properties. The new SVM model will assist petroleum exploration engineers to estimate various reservoir properties with better accuracy, leading to reduced exploration time and increased production.
Boosting for Mineral Prospectivity Modeling: A New GIS Toolbox
Natural Resources Research
With an increasing demand for raw materials, predictive models that support successful mineral exploration targeting are of great importance. We evaluated different machine learning techniques with an emphasis on boosting algorithms and implemented them in an ArcGIS toolbox. Performance was tested on an exploration dataset from the Iberian Pyrite Belt (IPB) with respect to accuracy, performance, stability, and robustness. Boosting algorithms are ensemble methods used in supervised learning for regression and classification. They combine weak classifiers, i.e., classifiers that perform slightly better than random guessing to obtain robust classifiers. Each time a weak learner is added; the learning set is reweighted to give more importance to misclassified samples. Our test area, the IPB, is one of the oldest mining districts in the world and hosts giant volcanichosted massive sulfide (VMS) deposits. The spatial density of ore deposits, as well as the size and tonnage, makes the area unique, and due to the high data availability and number of known deposits, well-suited for testing machine learning algorithms. We combined several geophysical datasets, as well as layers derived from geological maps as predictors of the presence or absence of VMS deposits. Boosting algorithms such as BrownBoost and Adaboost were tested and compared to Logistic Regression (LR), Random Forests (RF) and Support Vector machines (SVM) in several experiments. We found performance results relatively similar, especially to BrownBoost, which slightly outperformed LR and SVM with respective accuracies of 0.96 compared to 0.89 and 0.93. Data augmentation by perturbing deposit location led to a 7% improvement in results. Variations in the split ratio of training and test data led to a reduction in the accuracy of the prediction result with relative stability occurring at a critical point at around 26 training samples out of 130 total samples. When lower numbers of training data were introduced accuracy dropped significantly. In comparison with other machine learning methods, Adaboost is user-friendly due to relatively short training and prediction times, the low likelihood of overfitting and the reduced number of hyperparameters for optimization. Boosting algorithms gave high predictive accuracies, making them a potential data-driven alternative for regional scale and/or brownfields mineral exploration.
Minerals
Spatial prediction of orebody characteristics can often be challenging given the commonly complex geological structure of mineral deposits. For example, a high nugget effect can strongly impact variogram modelling. Geological complexity can be caused by the presence of structural geological discontinuities combined with numerous lithotypes, which may lead to underperformance of grade estimation with traditional kriging. Deep learning algorithms can be a practical alternative in addressing these issues since, in the neural network, calculation of experimental variograms is not necessary and nonlinearity can be captured globally by learning the underlying interrelationships present in the dataset. Five different methods are used to estimate an unsampled 2D dataset. The methods include the machine learning techniques Support Vector Regression (SVR) and Multi-Layer Perceptron (MLP) neural network; the conventional geostatistical methods Simple Kriging (SK) and Nearest Neighbourhood (NN)...