一种改进的基于区分矩阵的属性约简算法

A new algorithm of attribute reduction based on discernibly matrix

以属性在区分矩阵中出现的频率作为启发,对HORAFA算法做了一些改进。引入二进制可辨识矩阵,对区分函数进行简化,求出相对核。以相对核为基础,加入属性重要性最大的属性,直到不能再加。在此基础上,加上反向消除的过程,保证了算法的完整性。

The paper learns something from the frequency of attributes appeared in the discernibly matrix, and makes progress on the HORAFA. Lead binary discernibly matrix in, and make the discernibly function more simplify, and then find the core. The algorithm adds the most important attribute into the core until can’t add anymore. Then add a process of backward elimination to make sure the completeness of algorithm.