Introduction to mean field statistics

Catalog description

In this course, we will introduce some ideas and techniques (both rigorous and non-rigorous) originated from statistical physics along with their applications in statistics and machine learning, exemplified by a few fundamental statistical models such as the spiked matrix/tensor model and the linear model. Topics include the replica method, Bayesian informational theoretical limits, the approximate message passing algorithm, statistical-to-computational gaps, etc. Prerequisites: (15-781 or 10-601 ) and (36-705 or 36-725)