Variational Bayesian methods: Difference between revisions

Content deleted Content added
m Layout fix
Line 15:
=== Problem ===
 
In [[Calculus of variations|variational]] inference, the posterior distribution over a set of unobserved variables <math>\mathbf{Z} = \{Z_1 \dots Z_n\}</math> given some data <math>\mathbf{X}</math> is approximated by a so-called '''variational distribution''', <math>Q(\mathbf{Z}):</math>:
: <math>P(\mathbf{Z}\mid \mathbf{X}) \approx Q(\mathbf{Z}).</math>
 
The distribution <math>Q(\mathbf{Z})</math> is restricted to belong to a family of distributions of simpler form (e.g. a family of Gaussian distributions) than <math>P(\mathbf{Z}\mid \mathbf{X})</math>, selected with the intention of making <math>Q(\mathbf{Z})</math> similar to the true posterior, <math>P(\mathbf{Z}\mid \mathbf{X})</math>.