| 基于数据库信息构建贝叶斯网络的GA方法 |
| |
| 引用本文： | 王君圣,李敏强. 基于数据库信息构建贝叶斯网络的GA方法[J]. 系统工程与电子技术, 2000, 22(9): 54-57 |
| |
| 作者姓名： | 王君圣 李敏强 |
| |
| 作者单位： | 天津大学管理学院,300072 |
| |
| 基金项目： | 国家自然科学基金资助课题!(7940 0 0 1 3,699740 2 6) |
| |
| 摘 要： | 主要论述了以现实存在的大量数据库信息为基础构建贝叶斯网络 (BayesianNetworks ,BN)的主要问题。讨论了不确定性知识的图形表示方法 ,以及如何利用专家知识指导属性的排序与选择。鉴于网络结构的复杂度随论域中结点个数的增加呈指数上升 ,如何寻找最佳贝叶斯网络就成为NP -Hard难题 ,应用遗传算法 (GeneticAlgorithms,GA)给出了最佳贝叶斯网络设计的一种新方法 ,并进行了实例计算。
|
| 关 键 词： | 贝叶斯定理 网络 遗传 算法 数据库 |
| 修稿时间： | 1999-07-15 |
Constructing Bayesian Networks Based on Database by Using GA |
| |
| Wang Junsheng,Li Minqiang. Constructing Bayesian Networks Based on Database by Using GA[J]. System Engineering and Electronics, 2000, 22(9): 54-57 |
| |
| Authors: | Wang Junsheng Li Minqiang |
| |
| Abstract: | This paper discusses the constructing of Bayesian networks based on large databases. The description of non-deterministic knowledge in Bayesian networks is analyzed, and the calculation of the conditional probabilities of the nodes in the network is put forward. Since the optimal structure of Bayesian network is a typical NP-hard problem, we outline the K2 method and point out its limitation in search of the optimal solutions. Then we present the logic of using expert knowledge to guide the choice and selection of the father nodes in network design, and adopting genetic algorithms to get the global optimal network. Experiments reveal the effectiveness of the new method. |
| |
| Keywords: | Bayesian theorem Networks Genetic Algorithms Databases |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
