【摘要】 影响基于差别矩阵的属性约简算法效率的主要因素有计算U/C等价类和差别矩阵的大小。为了解决差别矩阵大小影响属性约简算法计算效率,分析了基于差别矩阵的属性约简算法中差别矩阵定义的不足,重新定义了一种压缩差别矩阵,删除差别矩阵中大量的空元素和相同元素,从而进一步减少了差别矩阵元素的个数,并设计基于压缩差别矩阵的属性约简算法。对UCI及其他数据库进行仿真,实验结果表明该算法具有高效性。更多还原

【Abstract】 Calculating the equivalence class U/C and the discernibility matrices size are the two main factors which affect the efficiency of the attribute reduction algorithm based on discernibility matrices.In order to solve the efficiency problem of calculating attribute reduction based on discernibility matrices,the disadvantages of attribution reduction algorithms were analyzed,and a condensing discernibility matrices was redefined.It greatly decreased the number of empty elements and same elements and improved the efficiency of algorithm for attribute reduction based on discernibility matrices.A new algorithm based on the condensing discernibility matrices was proposed.The simulation experiments for UCI and other databases show that the new algorithm is effective and efficient.更多还原