MATRL 289MT: ML FOUND: MAT SCI

University of California, Santa Barbara

Covers fundamental machine learning tools for materials science, including surrogate models for atomistic simulations, 3D image analysis, and experime ntal data analysis. It begins with probability theory and Bayesian inferenc e, followed by regularized regression, Gaussian processes, and neural netwo rks. Concepts of invariance and equivariance are introduced, with equivaria nt neural network architectures discussed. The course also covers ML method s for image analysis (e.g., variational autoencoders, diffusion models, GAN s) and natural language processing.