Topics in Bayesian Modeling and Computation

Graduate credit

This course provides basic concepts and several modern techniques of Bayesian modeling and computation. They include basic models, conjugate priors, and posterior computation, as well as techniques associated with complex models, such as hierarchical models, spatiotemporal models, and dynamical models. A substantial part of the course is devoted to computational algorithms based on Markov Chain Monte Carlo sampling for complex models. If time permits, we will also introduce advanced topics such as nonparametric Bayes, variational inference, Hamiltonian Monte Carlo, and sequential Monte Carlo techniques. Foundational topics will be discussed when appropriate, although they are not our primary focus in this course; such topics may include decision theoretic characterization of Bayesian inference and its relation to frequentist methods, de Finetti-type theorems and the existence of priors, objective prior distributions, and Bayesian model selection. Students taking the course under STATS 651 will have to complete an individual class project subject to the instructor's approval.