Title: Bayesian Inference in the Presence of Intractable Normalizing Functions
Abstract: Models with intractable normalizing functions arise frequently in statistics. Common examples of such models include Markov point processes for disease modeling, and exponential random graph models for social networks. Inference for these models is complicated because the normalizing "constants" of the distributions are actually functions of the parameters of interest. In Bayesian analysis this results in so-called doubly intractable distributions which pose significant computational challenges. My talk will: (i) provide an overview of algorithms for these problems, (ii) discuss practical computational issues, and (iii) outline a new two-stage emulation-based approach that is very fast for certain doubly intractable distributions. This talk is based on joint work with Jaewoo Park.