Menu
Home
People
Places
Arts
History
Plants & Animals
Science
Life & Culture
Technology
Reference.org
Variational message passing
open-in-new
<p>Variational message passing (VMP) is an <a href="/facts/Approximate_inference/IG9ogR9I">approximate inference</a> technique for continuous- or discrete-valued <a href="/facts/Bayesian_networks/Ry31vq6T">Bayesian networks</a>, with <a href="/facts/Conjugate_exponents/JbfcLdvB">conjugate-exponential</a> parents, developed by John Winn. VMP was developed as a means of generalizing the approximate <a href="/facts/Variational_Bayesian_methods/pFuMLBcU">variational methods</a> used by such techniques as <a href="/facts/Latent_Dirichlet_allocation/BUWlwKWt">latent Dirichlet allocation</a>, and works by updating an approximate distribution at each node through messages in the node's <a href="/facts/Markov_blanket/h3Ns4Nhj">Markov blanket</a>. </p>