Evidence Estimation for Bayesian Partially Observed MRFs
Y. Chen, Max Welling · UvA-DARE (University of Amsterdam) · 2013
Bayesian estimation in Markov random fields is very hard due to the intractability of the parti-tion function. The introduction of hidden units makes the situation even worse due to the pres-ence of potentially very many modes in the pos-terior distribution. For the first time we pro-pose a comprehensive procedure to address one of the Bayesian estimation problems, approxi-mating the evidence of partially observed MRFs based on the Laplace approximation. We also in-troduce a number of approximate MCMC-based methods for comparison but find that the Laplace approximation significantly outperforms these. 1