| 贝叶斯网络的参数学习研究 |
| |
| 引用本文： | 李晓毅,徐兆棣,孙笑微.贝叶斯网络的参数学习研究[J].沈阳农业大学学报,2007,38(1):125-128. |
| |
| 作者姓名： | 李晓毅 徐兆棣 孙笑微 |
| |
| 作者单位： | 1. 沈阳师范大学,数学与系统科学学院,沈阳,110034
2. 沈阳师范大学,软件学院,沈阳,110034 |
| |
| 摘 要： | 贝叶斯网络是数据采掘的一个非常有效的工具,它能够定性和定量地分析属性之间的依赖关系,进行概率推理。在给出贝叶斯网络相关概念后,讨论了贝叶斯网络在离散型、连续型和混合型3种情况下的参数学习计算公式;利用给定样本数据,对具有完整数据和不具有完整数据的贝叶斯网络,学习网络参数,更新网络变量原有的先验分布;从理论上给出两种不同类型下的参数学习公式。
|
| 关 键 词： | 贝叶斯网络 网络结构 参数学习 狄利克雷分布 数据采掘 |
| 文章编号： | 1000-1700（2007）01-0125-04 |
| 修稿时间： | 2006-03-05 |
Study on Parameter Learning of Bayesian Network |
| |
| LI Xiao-yi,XU Zhao-di,SUN Xiao-wei.Study on Parameter Learning of Bayesian Network[J].Journal of Shenyang Agricultural University,2007,38(1):125-128. |
| |
| Authors: | LI Xiao-yi XU Zhao-di SUN Xiao-wei |
| |
| Institution: | 1, College of Mathematics and Systems Science, 2,College of Software,Shenyang Normal University, Shenyang 110034, China |
| |
| Abstract: | Bayesian network as,a very useful tool in data mining,can provide qualitative and quantitative relationship between attributes and probability inference.This paper firstly reviewed the concept of Bayesian network,then discusseed the Bayesian network formula in cases of discrete,continuous and hybrid type.According to the data of the samples,for the Bayesian network with complete and incomplete data,network parameter was studied.The prior distribute of network variable was updated,and two different types of learning formula of parameter were given theoretically. |
| |
| Keywords: | Bayesian network network structure parameter learning Dirichlet distribute data mining |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
