基于互信息和Beam搜索的粗糙集属性约简算法

Rough set attribute reduction based on mutual information and Beam search

从属性集互信息的角度分析了粗糙集理论的属性约简问题.首先在互信息的基础上定义了一个新的属性子集的冗余性和协同能力度量——属性子集的冗余协同系数;然后将它作为属性约简度量,提出了基于Beam搜索的粗糙集属性约简算法.实验表明属性约简算法具有良好的运行效果.

The attribute reduction problem for rough set is analyzed by the mutual information of attribute set. Based on mutual information, the redundancy-synergy coefficient of attribute set, a novel measure for redundancy and synergistic ability of attribute set, is defined. A Beam search based attribute reduction algorithm for rough set is presented, where the redundancy-synergy coefficient is taken as the attribute reduction measure. Experiments show that the new algorithm yields satisfying attribute reduction results.