Universal Models for the Exponential Distribution

Author

Schmidt, Daniel F. ; Makalic, Enes

Author_Institution

Centre for MEGA Epidemiology, Univ. of Melbourne, Carlton, VIC

Volume

55

Issue

7

fYear

2009

fDate

7/1/2009 12:00:00 AM

Firstpage

3087

Lastpage

3090

Abstract

This paper considers the problem of constructing information theoretic universal models for data distributed according to the exponential distribution. The universal models examined include the sequential normalized maximum likelihood (SNML) code, conditional normalized maximum likelihood (CNML) code, the minimum message length (MML) code, and the Bayes mixture code (BMC). The CNML code yields a codelength identical to the Bayesian mixture code, and within O(1) of the MML codelength, with suitable data driven priors.

Keywords

codes; exponential distribution; Bayes mixture code; conditional normalized maximum likelihood code; exponential distribution; information theoretic universal models; minimum message length code; sequential normalized maximum likelihood code; Australia; Bayesian methods; Distributed computing; Exponential distribution; Integral equations; Maximum likelihood estimation; Minimax techniques; Parametric statistics; Predictive models; Statistical distributions; Minimum description length (MDL); minimum message length (MML); universal models;

fLanguage

English

Journal_Title

Information Theory, IEEE Transactions on

Publisher

ieee

ISSN

0018-9448

Type

jour

DOI

10.1109/TIT.2009.2018331

Filename

5075876