Probabilistic Machine Learning

Undergraduate

This course studies the theory and algorithms for probabilistic machine learning applications. With Bayesian inference as a motivation, we study algorithms for approximate posterior computation, including via optimization (variational inference) and via sampling (Markov chain Monte Carlo). We survey the relations between optimization and sampling problems. We study sampling algorithms as optimization algorithms on the space of probability distributions, and derive their theoretical guarantees. We survey modern generative processes, including score-based generative modeling, and discuss their theory, applications, and challenges. This course familiarizes students with basic techniques and current results in the literature, and prepares students for research in probabilistic machine learning. This course is designed for graduate students and advanced undergraduates who have strong mathematical foundation and previous knowledge of machine learning. Prerequisites: Introductory machine learning (CPSC 381 or CPSC 481 or S&DS 265 or S&DS 365); linear algebra (MATH 222); probability (S&DS 241); and calculus (MATH 120). 0 Course cr HTBA