ESTIMATION OF STATE PROBABILITIES USING THE MAXIMUM-ENTROPY PRINCIPLE

被引：15

作者：

BARD, Y

机构：

来源：

IBM JOURNAL OF RESEARCH AND DEVELOPMENT | 1980年 / 24卷 / 05期

关键词：

D O I：

10.1147/rd.245.0563

中图分类号：

TP3 [计算技术、计算机技术];

学科分类号：

0812 ;

摘要：

A simple method is derived for computing state probabilities of a system when the probabilities of certain aggregate states are known. The method is based on maximizing the system entropy. It is shown that the results obtained by the method satisfy certain assumptions on statistical independence between events. The method is applied to a problem arising in computer performance analysis.

引用

页码：563 / 569

页数：7

共 8 条

[1]

AGMON N, 1979, MAXIMUM ENTROPY FORM, P207

[2]

BARD Y, 1980, G3202130 IBM SCI CTR

[3]

CHAPMAN RW, 1977, RJ1930 IBM RES LAB R

[4] A STATISTICAL MECHANICAL APPROACH TO SYSTEMS ANALYSIS [J].

FERDINAN.AE .

IBM JOURNAL OF RESEARCH AND DEVELOPMENT, 1970, 14 (05) :539-&

[5]

GOKHALE DV, 1978, INFORMATION CONTINGE

[6]

KATZ A, 1967, PRINCIPLES STATISTIC

[7]

LEVINE RD, 1979, MAXIMUM ENTROPY FORM, P15

[8]

Shore J. E., 1978, AFIPS Conference Proceedings vol.47. 1978 National Computer Conference, P483

← 1 →