MATRL 289M: MCH LRN 4 MATRL SCI

University of California, Santa Barbara

Covers the fundamentals of machine learning tools used in materials science . Application areas include surrogate models for atomistic simulations as w ell as machine-learning tools for 3-dimensional image analysis and the anal ysis of experimental data. The course starts with an overview of probabilit y theory and Bayesian inference followed by treatments of regularized regre ssion, Gaussian process models and neural networks. Concepts of invariance and equivariance in the context of spatial transformations will be develope d and neural network architectures that are equivariant to translational an d rotational transformations will be analyzed.