Recognition: unknown
MCMC for doubly-intractable distributions
read the original abstract
Markov Chain Monte Carlo (MCMC) algorithms are routinely used to draw samples from distributions with intractable normalization constants. However, standard MCMC algorithms do not apply to doubly-intractable distributions in which there are additional parameter-dependent normalization terms; for example, the posterior over parameters of an undirected graphical model. An ingenious auxiliary-variable scheme (Moeller et al., 2004) offers a solution: exact sampling (Propp and Wilson, 1996) is used to sample from a Metropolis-Hastings proposal for which the acceptance probability is tractable. Unfortunately the acceptance probability of these expensive updates can be low. This paper provides a generalization of Moeller et al. (2004) and a new MCMC algorithm, which obtains better acceptance probabilities for the same amount of exact sampling, and removes the need to estimate model parameters before sampling begins.
This paper has not been read by Pith yet.
Forward citations
Cited by 2 Pith papers
-
Penalized Likelihood for Dyadic Network Formation Models with Degree Heterogeneity
Penalized likelihood resolves non-existence of MLE and incidental-parameter bias in network models with degree heterogeneity while allowing sparse networks and providing asymptotic guarantees.
-
A unifying view of contrastive learning, importance sampling, and bridge sampling for energy-based models
A unified framework shows that NCE, RLR, MIS, and bridge sampling are equivalent under specific conditions for energy-based models, enabling new estimators.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.