Home // ADVCOMP 2021, The Fifteenth International Conference on Advanced Engineering Computing and Applications in Sciences // View article

A Novel Application of Machine Learning to a New SEM Silicate Mineral Dataset

Authors:
Benjamin Parfitt
Robert Welch

Keywords: machine learning, ensemble learning, mineralogy, silicates.

Abstract:
Machine Learning (ML) continues to find applications in the geosciences, specifically in the classification of minerals from spectral or elemental data. We begin by exploring the use of four different methods for classification of elemental mineral samples from Scanning Electron Microscopy (SEM) and microprobe analysis in terms of structure, group, and subgroup. We create the most extensive silicate mineral group and subgroup classifiers available to the best of our knowledge, and achieve precision and recall values as high as the current state-of-the-art methods, which cover fewer groups and subgroups. Finally, we attempt to leverage the knowledge of structural families to improve classification performance on mineral groups, and reapply this process to improve performance on mineral subgroups. The train, test, validation split of data used in this paper will be posted online, along with the code and a webpage called MINdicator where anyone can use the new models easily.

Pages: 11 to 17

Copyright: Copyright (c) IARIA, 2021

Publication date: October 3, 2021

Published in: conference

ISSN: 2308-4499

ISBN: 978-1-61208-887-7

Location: Barcelona, Spain

Dates: from October 3, 2021 to October 7, 2021