A0773
Title: A correlated pseudo-marginal approach to doubly intractable problems
Authors: Matias Quiroz - University of Technology Sydney (Australia) [presenting]
Robert Kohn - University of New South Wales (Australia)
Scott Sisson - University of New South Wales (Austria)
Yu Yang - University of New South Wales (Australia)
Abstract: Doubly intractable models are encountered in several fields, e.g. social networks, ecology, and epidemiology. Inference for such models requires the evaluation of a likelihood function, whose normalizing function depends on the model parameters and is typically computationally intractable. A signed pseudo-marginal Metropolis-Hastings (PMMH) algorithm is proposed with an unbiased block-Poisson estimator to sample from the posterior distribution of doubly intractable models. The advantage of the estimator over previous approaches is that its form is ideal for correlated pseudo-marginal methods, which are well known to increase sampling efficiency dramatically. Moreover, analytically derived heuristic guidelines are developed for optimally tuning the hyperparameters of the estimator.
