一个基于差别矩阵的属性约简改进算法

An Updated Algorithm for Attribute Reduction Based on Discernibility Matrix

介绍了决策表的基于差别矩阵的属性约简方法,针对AM-RASR约简算法的不足,将叶东毅的对不相容决策表的求核思想融合进来,提出一个改进的算法,新算法通过在扫描数据过程中用冲突对象对差别集进行修正,可以有效地减少数据的存储量和计算量,并能适用于不相容决策表.最后通过一个UCI数据集的实验说明改进的算法是有效的.

Attribute reduction is one of the most important content of the Rough Set Theory. This paper analyzes attribute reduction for decision table based on discernibility matrix and YE＇s thought for calculating the core of inconsistent decision tables. As AM-RASR algorithm needs too much storing capacity and doesn＇ t support inconsistent decision tables, and YE＇s method needs much more computing, this paper presents an improved algorithm. Through revising discernibility collection with conflicting element when scanning data, which can cut down the original algorithm＇s computing and storing capacity, and can be applied in inconsistent decision tables. Fianlly, through an experiment with UCI data sets, the paper explains the effectiveness of the improved algorithm.