基于分明矩阵的属性约简启发式算法

Heuristic algorithm to rough set attribute reduction based on discernibility matrix

提出了基于分明矩阵的启发式知识约简算法。该算法以分明矩阵中属性出现的频率作为启发信息,通过构造新的决策表,每次选取出现个数最多的属性,直到选取的属性能够保持原决策表的分类能力,此时得到的集合即是一个约简。试验结果表明,该算法在大多数情况下都能够找到最小约简或令人满意的次优解。

An algorithm to heuristic knowledge reduction based on discernibility matrix is proposed. According to the frequency of attribute in discernibility matrix, a new information system is designed and the maximum attribute every time is selected, until the selected attribute can discern information system, then all selected attribute is a reduction. The practical results show that in the most cases the minimum or a feasible solution reduction is obtained using the proposed method.