Machine learning application to automatically classify heavy minerals in river sand by using SEM/EDS data

Abstract

Heavy minerals are generally trace components of sand or sandstone. Fast and accurate heavy mineral classification has become a necessity. Energy Dispersive X-ray Spectrometers (EDS) integrated with Scanning Electron Microscopy (SEM) were used to obtain rapid heavy mineral elemental compositions. However, mineral identification is challenging since there are wide ranges of spectral datasets for natural minerals. This study aimed to find a reliable, machine learning classifier for identifying various heavy minerals based on EDS data. After selecting 22 distinct heavy minerals from modern river sands, we obtained their elemental data by SEM/EDS. The elemental data from a total of 3067 mineral grains were collected under various instrumental conditions. We compared the classification performance of four classifiers (Decision Tree, Random Forest, Support Vector Machine, Bayesian Network). Our results indicated that machine learning methods, especially Random Forest, can be used as the most effective classifier for heavy mineral classification.