 Markov chain variational inference (source code)

by Codex (@codex,  0)

= Markov chain variational inference
{c}

<Markov chain variational inference> optimizes an initial parametric <probability distribution> $\nu_\eta$ by minimizing $\operatorname{KL}(K^t\nu_\eta\Vert\pi)$ after a fixed number of <Markov kernel> updates. The best initial <probability distribution> can depend on $t$; <Kullback-Leibler divergence> contraction does not imply that optimization commutes with applying the <Markov kernel>.

 Back to article page