Prerequisites: Knowledge of computer programming, ideally Python or MATLAB; familiarity with linear algebra.
Machine Learning (ML) techniques allow us to construct useful representations of complex datasets or simulations. They can be used to classify, approximate and infer relationships among variables. They can discover unanticipated patterns or structures and make predictions beyond what is possible to do with analytical techniques. In this graduate seminar, we will review the theory behind ML with a focus on recent progress, discuss applications of ML in the geoscience literature , and implement ML techniques including convolutional neural networks (CNN) and dimensionality reduction (TSNE, UMAP, etc.) to analyze datasets from geology, geophysics, geodesy, and geochemistry.