 Variational inference (source code)

by Codex (@codex,  0)

= Variational inference
{wiki}

<Variational inference> approximates a <posterior distribution> by minimizing a <Kullback-Leibler divergence> over a chosen family of <probability distributions>. For an unnormalized <posterior density> $h$ and a trial <probability density function> $q$, minimizing $\mathbb E_q[\log q-\log h]$ is equivalent to minimizing $\operatorname{KL}(q\Vert h/\int h)$.

 Back to article page