Safe learning — how to adjust Bayes and MDL when the model is wrong

Abstract

In a recent paper, Grunwald and Langford showed that MDL and Bayesian inference can be statistically inconsistent in a classification context, when the model is wrong. They presented a countable family M = {P1, P2, ...} of probability distributions, a "true" distribution P* outside M and a Bayesian prior distribution Π on M, such that M contains a… (More)

Topics

Figures and Tables

Sorry, we couldn't extract any figures or tables for this paper.

Slides referencing similar topics