摘要
阐明了变精度粗糙集模型中,经典粗糙集模型分类质量、相对正域、决策类下近似不再具有非单调递减特征,在约简过程中分类质量和相对正域会出现跳跃现象,约简过程具有不稳定性;但决策类下近似不会出现跳跃现象,可以得到稳定的约简过程;并且三者之间打破了在经典粗糙集模型中的等价性,需要针对三者分别建立模型,使属性约简变得多样化。
The paper illuminated important differences between classical rough sets and variable precision rough sets. In variable precision rough sets, there is no non-monotonic decreasing characteristics in quality of classification and relative positiveregion and bounce phenomena maybe occur, so the process of reduction is not stabile. Though the change of lower approximation also break non-monotonic decreasing principle, bounce phenomena don' t occur and stabile process of reduction can be gotten. The equivalence among quality of classification, relative positive-region and lower approximation is not exit in variable precision rough sets, attributes reduction becomes multiform.
出处
《计算机应用研究》
CSCD
北大核心
2007年第7期10-12,15,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(60474041)
湖南省自然科学基金资助项目(06JJ20075)
关键词
变精度粗糙集
非单调递减
跳跃现象
variable precision rough sets
non-monotonic decreasing
bounce phenomena