University of Cambridge > Talks.cam > Statistics Reading Group > Variational methods continued

  Browse A-Z

Log in

University Account External (via Google)

🔐 Log In

Information on

Subscribing to talksdetails Finding a talkdetails Adding a talkdetails Disseminating talksdetails Help and Documentationdetails

ℹī¸ Information

Variational methods continued

Download to your calendar using vCal

If you have a question about this talk, please contact Richard Samworth .

I will continue Silvia’s discussion of variational methods, and try to answer some of the questions raised during her talk. We saw how mean field inference can give a lower bound on the log partition function Z. I will describe a general message passing framework based on alpha divergences, which has mean field and expectation propagation (EP, a generalisation of belief propagation for continuous random variables) as special cases, as well as tree reweighted belief propagation, which can give an upper bound on Z. EP will be shown to greatly outperform both Gibbs sampling and mean field on certain problems. If there is enough time I will present some work on choosing the optimal tree to approximate a loopy graph or even the right distribution over trees.

This talk is part of the Statistics Reading Group series.

This talk is included in these lists:

Note that ex-directory lists are not shown.

Other lists

Hybrid: Seminar room, SPRI & Zoom - email organiser for details CMS Computer Room, Faculty of Music (11 West Road, Cambridge, CB3 9DP); Zoom meeting: https://cam-ac-uk.zoom.us/j/95274848632?pwd=M1hWOVVyeXZZOHFFT0YrdDF6WnRqZz09; Meeting ID: 952 7484 8632; Passcode: 677829 W2.02, Cambridge Judge Business School

Other talks

Title to be confirmed Using A Function-Centric Lens to Re-consider Regularisation, Representation Transfer and Geometric Properties of Neural Networks Cambridge RNA Club - IN PERSON Studying viruses to understand what makes pandemic viruses special and how innate immunity works Title TBC From Surrogate Solvers to PINNs and Beyond Classical Methods
 

Š 2006-2025 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity