欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
现有的贝叶斯网络增量学习方法忽略结构与参数变化的特点和内在联系,往往会降低更新后贝叶斯网络的可靠性。针对这一情况,提出了贝叶斯网络结构与参数变化并不同步,参数变化到一定程度将引起结构变化,并基于这种不同步性,给出了一种是否进行结构更新的判别方法,以及结构与参数更新的实现算法,实验结果显示,这种增量学习方法更加合理和可行。
英文摘要：
At present, the characteristic and immanent relationship between the structure and parameters change are ignored in Bayesian network incremental learning. And the reliability of updated Bayesian network maybe fall. The change of Bayesian network structure and parameters is not synchronous. The change of structure will happen when parameters alter to a certain extent. A distinguishing criterion of whether revising structure was developed. The arithmetic of regulating structure and parameters was proposed. Experimental results show that this method of Bayesian network incremental learning is more rational and feasible.
同期刊论文项目
同项目期刊论文
期刊信息
位置：