Topics in Flow-Based Generative Modeling and Optimal Transport

Graduate / Professional

Generative AI has had an immense impact on many scientific disciplines over the last several years. A driving force has been the development of scalable algorithms that allow practitioners to “flow” from simple distributions to complex ones. In this course, we explore the mathematical and algorithmic foundations of generative models through the lens of flow-based transport. Topics include flow matching, stochastic interpolants, diffusion models, optimal transport maps, Schrödinger bridges, and their applications and extensions. Prerequisites: probability theory, multivariate calculus, and linear algebra, some programming, and permission of the instructor. M 9:25am-11:20am