Prune Sampling is a new MCMC method for discrete Bayesian networks that provably converges to the posterior distribution even under determinism, though it trails standard methods on large networks.
Stochastic relaxation, gibbs distributions, and the bayesian restoration of images
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.CO 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Prune Sampling: a MCMC inference technique for discrete and deterministic Bayesian networks
Prune Sampling is a new MCMC method for discrete Bayesian networks that provably converges to the posterior distribution even under determinism, though it trails standard methods on large networks.