Learning Parameters in Canonical Models Using Weighted Least Squares

被引：0

作者：

Nowak, Krzysztof ^{[1
,3
]}

Druzdzel, Marek J. ^{[1
,2
]}

机构：

[1] Bialystok Tech Univ, Bialystok, Poland

[2] Sch Informat Sci, Pittsburgh, PA USA

[3] European Space Agcy, NL-2200 AG Noordwijk, Netherlands

来源：

PROBABILISTIC GRAPHICAL MODELS | 2014年 / 8754卷

关键词：

Bayesian networks; canonical models; noisy-MAX gates; parameter learning; weighted least squares;

D O I：

暂无

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

We propose a novel approach to learning parameters of canonical models from small data sets using a concept employed in regression analysis: weighted least squares method. We assess the performance of our method experimentally and show that it typically outperforms simple methods used in the literature in terms of accuracy of the learned conditional probability distributions.

引用

页码：366 / 381

页数：16

共 13 条

[1] Diez F. J., 2006, TECHNICAL REPORT
[2] Diez F.J., 1993, PROC 9 C UNCERTAINTY, P99, DOI DOI 10.1016/B978-1-4832-1451-1.50016-0
[3] FRIEDMAN N, 1999, LEARNING GRAPHICAL M, P421
[4] On choosing and bounding probability metrics
Gibbs, AL
Su, FE
[J]. INTERNATIONAL STATISTICAL REVIEW, 2002, 70 (03) : 419 - 435
[5] Heckerman D., 1998, Learning in Graphical Models
[6] Henrion M., 1987, ART INT 3 ANN C UNC, P161
[7] Learning Bayesian network parameters from small data sets: application of Noisy-OR gates
Onisko, A
Druzdzel, MJ
Wasyluk, H
[J]. INTERNATIONAL JOURNAL OF APPROXIMATE REASONING, 2001, 27 (02) : 165 - 182
[8] Impact of precision of Bayesian network parameters on accuracy of medical diagnostic systems
Onisko, Agnieszka
Druzdzel, Marek J.
[J]. ARTIFICIAL INTELLIGENCE IN MEDICINE, 2013, 57 (03) : 197 - 206
[9] Pearl J., 1988, PROBABILISTIC REASON, DOI DOI 10.1016/C2009-0-27609-4
[10] SRINIVAS S, 1993, P 9 C UNC ART INT, P208, DOI DOI 10.1016/B978-1-4832-1451-1.50030-5

← 1 2 →