DocumentCode :
752069
Title :
Large-scale typicality of Markov sample paths and consistency of MDL order estimators
Author :
Csiszár, Imre
Author_Institution :
A. Renyi Inst. of Math., Hungarian Acad. of Sci., Budapest, Hungary
Volume :
48
Issue :
6
fYear :
2002
fDate :
6/1/2002 12:00:00 AM
Firstpage :
1616
Lastpage :
1628
Abstract :
For Markov chains of arbitrary order, with finite alphabet A, almost sure sense limit theorems are proved on relative frequencies of k-blocks, and of symbols preceded by a given k-block, when k is permitted to grow as the sample size n grows. As-an application, the-consistency of two kinds of minimum description length (MDL) Markov order estimators is proved, with upper bound o(log n), respectively, α log n with α < 1/log |A|, on the permissible value of the estimated order. It was shown by Csiszar and Shields (see Ann. Statist., vol.28, p.1601-1619, 2000) that in the absence of any bound, or with bound α log n with large α consistency fails
Keywords :
Markov processes; encoding; maximum likelihood estimation; Krichevsky-Trofimov distribution; MDL order estimators; Markov chains; Markov order estimators; Markov sample paths; almost sure sense limit theorems; codeword; consistency; finite alphabet; large-scale typicality; minimum description length; normalized maximum likelihood coding distribution; sample size; stochastic process; uniform distribution; upper bound; Estimation theory; Frequency estimation; H infinity control; Information theory; Large-scale systems; Mathematics; Maximum likelihood estimation; Stochastic processes; Upper bound;
fLanguage :
English
Journal_Title :
Information Theory, IEEE Transactions on
Publisher :
ieee
ISSN :
0018-9448
Type :
jour
DOI :
10.1109/TIT.2002.1003842
Filename :
1003842
Link To Document :
بازگشت