On the estimation of Markov random field parameters

被引：12

作者：

Borges, CF ^{[1
]}

机构：

[1] USN, Postgrad Sch, Monterey, CA 93943 USA

来源：

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 1999年 / 21卷 / 03期

关键词：

Markov random fields; parameter estimation;

D O I：

10.1109/34.754587

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

We examine the histogram method proposed in [1] for estimating the parameters associated with a Markov random field. This method relies on the estimation of the local interaction sums from histogram data. We derive an estimator for these quantities that is optimal in a well-defined sense. Furthermore. we show that the final step of the histogram method, the solution of a least-squares problem. can be done substantially faster than one might expect it no equation culling is used. We also examine the use of weighted least-squares and see that this seems to lead to better estimates even with smalt amounts of data.

引用

页码：216 / 224

页数：9