论文部分内容阅读
知识约简是粗糙集理论重要研究内容之一.目前,通常采用差别矩阵来计算不一致决策表的分布约简、最大分布约简和分配约简,可以获得所有约简,但算法复杂度较高,在较大数据集下非常耗时.针对不一致决策表,提出不一致决策表的一致化决策表转化算法,将计算原不一致决策表的约简转化为计算一致化决策表的约简.给出了5种一致化决策表约简的定义,探讨了各种约简结果之间的关系,利用相对不可辨识的对象对构建了高效的差别矩阵知识约简算法.理论分析和实验结果表明,本文所提出的算法能够有效地减少计算时间,适合处理大数据集.
Knowledge reduction is one of the most important research topics in rough set theory.At present, the discernibility matrix is usually used to calculate the distributive reduction, maximum distributive reduction and distributive reduction of disagreement decision tables, and all the reductions can be obtained, but the algorithm is more complex , Which is very time-consuming under a large data set.For the inconsistent decision table, a uniform decision table transformation algorithm of inconsistent decision table is proposed, and the reduction of the original inconsistent decision table is converted into the reduction of the consistent decision table. The definition of five kinds of consistent decision table reduction is discussed and the relationship between various reduction results is explored.The efficient discriminative matrix knowledge reduction algorithm is constructed by using relatively unidentifiable objects.The theoretical analysis and experimental results show that this article The proposed algorithm can effectively reduce the computational time, suitable for processing large data sets.