Open Collections

Abstract

This work looks at fitting probabilistic graphical models to data when the structure is not known. The main tool to do this is L₁-regularization and the more general group L₁-regularization. We describe limited-memory quasi-Newton methods to solve optimization problems with these types of regularizers, and we examine learning directed acyclic graphical models with L₁-regularization, learning undirected graphical models with group L₁-regularization, and learning hierarchical log-linear models with overlapping group L₁-regularization.