Variational Inference: An Algorithm-Centric Perspective
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact Xianda Sun.
This tutorial will introduce variational inference (VI), a family of algorithms for approximate Bayesian inference. We will begin by motivating the need for VI; when is it useful in the age of powerful Markov chain Monte Carlo algorithms such as NUTS ? The tutorial will
then take a top-down approach, introducing the abstract setup of variational inference. The effect of each component of the setup, such as the choice of variational family and divergence measure, will be illustrated. The rest of the tutorial will walk through the historical evolution of VI algorithms, but on a conceptual level, focusing on intuitions. For instance, what are the key ideas of a particular algorithm? What are its requirements and assumptions? What makes it
effective, and what are its limitations?
This talk is part of the Machine Learning Reading Group @ CUED series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|