We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Suboptimal behavior of Bayes and MDL in classification under misspecification.
- Authors
Peter Grünwald; John Langford
- Abstract
Abstract??We show that forms of Bayesian and MDL inference that are often applied to classification problems can beinconsistent. This means that there exists a learning problem such that for all amounts of data the generalization errors of the MDL classifier and the Bayes classifier relative to the Bayesian posterior both remain bounded away from the smallest achievable generalization error. From a Bayesian point of view, the result can be reinterpreted as saying that Bayesian inference can be inconsistent under misspecification, even for countably infinite models. We extensively discuss the result from both a Bayesian and an MDL perspective.
- Subjects
BAYESIAN analysis; STATISTICAL decision making; ALGORITHMS; ERROR analysis in mathematics
- Publication
Machine Learning, 2007, Vol 66, Issue 2-3, p119
- ISSN
0885-6125
- Publication type
Article
- DOI
10.1007/s10994-007-0716-7