PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

On the Bahadur-efficient Testing of Uniformity by means of Entropy
Peter Harremoes and Igor Vajda
IEEE Transactions on Information Theory Volume 54, Number 1, pp. 321-331, 2008. ISSN 0018-9448

Abstract

This paper compares the power divergence statistics of order $\alpha >1$ with the information divergence statistics in the problem of testing the uniformity of a dsitribution. In this problem the information divergence statistics is equivalent to the entropy statistics. Extending some previously established results about information diagrams, it is proved that the information divergence statistics in this problem is more efficient in hte Bahadur sense than any other power divergence statistics of order $\alpha > 1$. This means that the entropy provides in this sense the most efficient way of characterizing the uniform distribution.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Article
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
ID Code:3446
Deposited By:Peter Harremoes
Deposited On:11 February 2008