摘要
通过引入知识粒度的概念,对信息系统中属性的重要度进行了定义,并以属性重要度为启发式信息,进行粗集的属性约简.在构造决策树的过程中,基于粗集的理论运用了加权平均粗糙度的概念,并将其作为选择分离属性的标准.将这种联合粗集与决策树的模型应用到雷达信号识别中,经实验证明,用该方法构造的决策树复杂性低,且能有效提高分类效果.
The significance of attribute in information s^tem is defined by applying the concept of granulation; then the significance of attribute can be treated as the heuristic information to attribute reduction in rough set. In the process of constructing a decision tree, weighted mean roughness, a new concept based on rough set theory which is regarded as the criteria for choosing attributes is applied. The model which combines rough set and decision tree is used in radar signal recognition and the experiments show that the decision tree constructed in this paper is simpler in structure, and can improve the efficiency of classification.
出处
《微电子学与计算机》
CSCD
北大核心
2008年第12期13-16,共4页
Microelectronics & Computer
基金
国家"八六三"计划项目(2005AA775020)
关键词
粒度
属性约简
粗集
决策树
加权平均粗糙度
granulation
attribute reduction
rough set
decision tree
weighted mean roughness
