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：
提出一种新的学习无约束贝叶斯网络分类器的算法（RE-BNC）.该算法基于粗糙集理论,在保证分类精度不变的前提下,先对冗余属性变量进行约简,降低属性变量维数,然后构建一个无约束优化模型用来学习较好的初始种群,降低搜索空间,再结合进化算法学习分类器的网络结构.与其他常见的8种分类器算法相比较,实验结果表明该算法设计合理,且分类效果较好.
英文摘要：
A new algorithm for learning Bayesian network classifiers based on rough set approach and evolutionary algorithm（RE-BNC） was proposed.Firstly,to reduce the dimensions of the attribute variables,redundant attributes of the to-be-classified data set were removed according to the rough set theory,and the accuracy was nondecreasing.Then an unconstrained optimization model was constructed for learning a better initial population and reducing the search space.At last,the structure of the classifier was learned by taking advantage of the evolutionary algorithm.Compared with other 8 kinds of classifier algorithms,the results showed that RE-BNC was reasonable and had higher accuracy.
同期刊论文项目
同项目期刊论文
期刊信息
位置：