Abstract
This paper considers a generalization of the connection between Jeffreys prior and the Kullback-Leibler divergence as a procedure for generating a wide class of invariant priors of which Jeffreys prior is only one. By viewing Jeffreys' approach as a special case of a more general procedure, we can see that the choice of Jeffreys prior entails both parametrization invariance and sample space invariance. This general procedure also provides a link between distributional discrepancy measures and Haar measure.
Original language | English (US) |
---|---|
Pages (from-to) | 169-179 |
Number of pages | 11 |
Journal | Journal of Statistical Planning and Inference |
Volume | 37 |
Issue number | 2 |
DOIs | |
State | Published - Nov 1993 |
Externally published | Yes |
Keywords
- Divergence measures
- Haar measure, information measures
- Jeffreys prior
- invariant priors
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty
- Applied Mathematics