PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

distribution-dependent PAC-Bayes priors
Francois Laviolette, Guy Lever and John Shawe-Taylor
In: Algorithmic Learning Theory, ALT 2010, Oct 2010, Canberra, Australia.

Abstract

We develop the idea that the PAC-Bayes prior can be informed by the data-generating distribution. We prove sharp bounds for an existing framework, and develop insights into function class complexity in this model and suggest means of controlling it with new algorithms. In particular we consider controlling capacity with respect to the unknown geometry of the data-generating distribution. We finally extend this localization to more practical learning methods.

EPrint Type:Conference or Workshop Item (Paper)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
ID Code:9237
Deposited By:Guy Lever
Deposited On:21 February 2012