Learning Mixed Graphical Models

We consider the problem of learning the structure of a pairwise graphical model over continuous and discrete variables. We present a new pairwise model for graphical models with both continuous and discrete variables. We estimate the parameters of this model by approximating the likelihood with the pseudolikelihood and regularizing with group-sparsity penalties. We also consider a conditional model that incorporates features. Two algorithms for solving the optimization problem are presented. The proposed models are compared with competing methods on synthetic data and a survey dataset.