Divergence measures and latent variable models
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Ryan Prescott Adams.
Alpha-divergences have attracted recent attention as a measure of divergence for generalizing algorithms for approximate inference (like Variational Bayes and Expectation Propagation) into a unified framework. This talk explores the use and effect of such divergence measures under mutivariate mixtures of Gaussians, an example of a simple non-trivial latent variable model. In particular, symmetry-breaking, Ockam-hills for different measures of divergence, behaviour under multiple modes, and practical results are discussed.
This talk is part of the Inference Group series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|