Natural gradient descent and variational inference

Add to your list(s) Download to your calendar using vCal

• Mark van der Wilk; Alex Matthews
• Thursday 18 February 2016, 14:30-16:00
• Engineering Department, CBL Room 438.

If you have a question about this talk, please contact Yingzhen Li.

We will cover the following topics, time allowing.

1) Knuth covariance 2) A short introduction to Riemannian geometry. 3) The elegant properties of the Fisher-Rao metric and the connection to the VB update. 4) The paper “Approximate Riemannian Conjugate Gradient Learning for Fixed-Form Variational Bayes” JMLR 2010 . Honkela, Raiko, Kuusela, Tornio, Karhunen. 5) The paper “Stochastic variational inference” JMLR 2013 Hoffman, Blei, Wang, Paisley. 6) Time allowing. The paper “Gaussian processes for Big Data” Hensman, Fusi, Lawrence.

There is no need to do background reading, but those wanting to get a flavour of the talk should look at the natural gradient parts of 4) and 5).

This talk is part of the Machine Learning Reading Group @ CUED series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Cambridge Centre for Data-Driven Discovery (C2D3)
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge University Engineering Department Talks
• Cambridge talks
• Centre for Smart Infrastructure & Construction
• Chris Davis' list
• Computational Continuum Mechanics Group Seminars
• Engineering Department, CBL Room 438
• Featured lists
• Guy Emerson's list
• Hanchen DaDaDash
• Inference Group Journal Clubs
• Inference Group Summary
• Information Engineering Division seminar list
• Interested Talks
• ML
• Machine Learning Reading Group
• Machine Learning Reading Group @ CUED
• Machine Learning Summary
• Quantum Matter Journal Club
• Required lists for MLG
• School of Technology
• Simon Baker's List
• TQS Journal Clubs
• Trust & Technology Initiative - interesting events
• bld31
• custom
• ndk22's list
• ob366-ai4er
• rp587
• yk373's list

Note that ex-directory lists are not shown.