MINIMUM CROSS-ENTROPY ANALYSIS WITH ENTROPY-TYPE CONSTRAINTS

被引：4

作者：

FANG, SC

PETERSON, EL

RAJASEKERA, JR

机构：

[1] N CAROLINA STATE UNIV, RALEIGH, NC 27695 USA

[2] AT&T BELL LABS, ENGN RES CTR, PRINCETON, NJ 08540 USA

来源：

JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS | 1992年 / 39卷 / 02期

关键词：

ENTROPY OPTIMIZATION; INFORMATION THEORY; GEOMETRIC PROGRAMMING; PROBABILITY INFERENCE;

D O I：

10.1016/0377-0427(92)90127-J

中图分类号：

O29 [应用数学];

学科分类号：

070104 ;

摘要：

In this paper, we study the cross-entropy optimization problem with entropy-type constraints. A simple geometric inequality is used to derive its dual problem and to show the "strong duality theorem". We found this geometric dual is a computationally attractive canonical program that is always consistent. A "dual-to-primal" conversion formula and a "dual perturbation" algorithm are also derived for computations.

引用

页码：165 / 178

页数：14

共 21 条

[1] THE ROLE OF DUALITY IN OPTIMIZATION PROBLEMS INVOLVING ENTROPY FUNCTIONALS WITH APPLICATIONS TO INFORMATION-THEORY [J].