Limits of information, Markov chains, and projection

被引：15

作者：

Barron, AR ^{[1
]}

机构：

[1] Yale Univ, Dept Stat, New Haven, CT 06520 USA

来源：

2000 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY, PROCEEDINGS | 2000年

关键词：

D O I：

10.1109/ISIT.2000.866315

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 [计算机科学与技术];

摘要：

The chain rule of information shows that log densities form Cauchy sequences, convergent in L-1, proving information limits, Markov chain convergence, and existence of information projections.

引用

页码：25 / 25

页数：1

共 7 条

[1]

THE STRONG ERGODIC THEOREM FOR DENSITIES - GENERALIZED SHANNON-MCMILLAN-BREIMAN THEOREM [J].

BARRON, AR .

ANNALS OF PROBABILITY, 1985, 13 (04) :1292-1303

[2]

COMPETITIVE OPTIMALITY OF LOGARITHMIC INVESTMENT [J].

BELL, RM ;

COVER, TM .

MATHEMATICS OF OPERATIONS RESEARCH, 1980, 5 (02) :161-166

[3]

SANOV PROPERTY, GENERALIZED I-PROJECTION AND A CONDITIONAL LIMIT-THEOREM [J].

CSISZAR, I .

ANNALS OF PROBABILITY, 1984, 12 (03) :768-793

[4]

FRITZ J, 1973, T 6 PROGR C INF THEO

[5]

KIEFFER JC, 1991, ALMOST SURE CONVERGE, V2, P151

[6]

LI JQ, 1999, ESTIMATION MIXTURE M

[7]

TOPSOE F, 1979, KYBERNETIKA, V15, P8

← 1 →