GENERAL ENTROPY CRITERIA FOR INVERSE PROBLEMS, WITH APPLICATIONS TO DATA-COMPRESSION, PATTERN-CLASSIFICATION, AND CLUSTER-ANALYSIS

被引：60

作者：

JONES, LK

BYRNE, CL

机构：

[1] Department of Mathematics University of Lowell, Lowell, MA

来源：

IEEE TRANSACTIONS ON INFORMATION THEORY | 1990年 / 36卷 / 01期

关键词：

D O I：

10.1109/18.50370

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Minimum distance approaches are considered for the reconstruction of a real function from finitely many linear functional values. An optimal class of distances satisfying an orthogonality condition analogous to that enjoyed by linear projections in Hilbert space is derived. These optimal distances are related to measures of distance between probability distributions recently introduced by Rao and Nayak and possess the geometric properties of cross entropy useful in speech and image compression, pattern classification, and cluster analysis. Several examples from spectrum estimation and image processing are discussed. © 1990 IEEE

引用

页码：23 / 30

页数：8

共 22 条

[1]

BURG JP, 1967, 37TH ANN M SOC EXPL

[2]

Charnes A., 1978, Mathematische Operationsforschung und Statistik, Series Optimization, V9, P21, DOI 10.1080/02331937808842461

[3] I-DIVERGENCE GEOMETRY OF PROBABILITY DISTRIBUTIONS AND MINIMIZATION PROBLEMS [J].