基于下近似单调的变精度粗糙集属性约简方法  

Attribute reduction method in variable precision rough set based on lower approximate monotonicity

作　　者：何俊红[1] 

机构地区：[1]宝鸡文理学院数学系,陕西宝鸡721013

出　　处：《计算机工程与应用》2015年第15期138-142,共5页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61402015);宝鸡文理学院院级重点项目(No.13JG11)

摘　　要：单调性在经典粗糙集属性约简过程中发挥着重要的作用。然而,在一些扩展模型中该单调性质并不存在,如变精度粗糙集模型。针对该问题,提出了变精度粗糙集模型中下近似单调约简的定义,下近似单调约简算法打破了传统意义上属性约简保持下近似不发生变化的局限性,认为属性约简可以追求下近似集尽可能增大。同时给出了求得该约简的属性约简方法。实验结果表明,相较于下近似保持约简算法,下近似单调约简算法求得的约简不仅增加了正域规则数目也减少了边界域规则数目,而且提高了数据的分类精度。由此可见,下近似单调约简算法增加了由正域表示的确定性,同时降低了由边界域带来的不确定性。It is well-known that, the monotonicity plays an important role in attribute reduction of classical rough set. However, such property does not always hold in some generalization models. Variable precision rough set is a typical example.From this point of view, the definition of lower approximate monotonicity attribute reduction is presented in variable precision rough set model. In lower approximate monotonicity attribute reduction, the decision maker prefers to increase the lower approximate set rather than preserving the lower approximate set unchanged. The attribute reduction approach is also given to compute the reduct. The experiment results show that by comparing with lower approximate preservation reduct, the lower approximate monotonicity reduction not only increases the number of positive rules, decreases the number of boundary rules, but also increases the classification accuracy. It follows that, the lower approximate monotonicity reduction increases the certainties which are expressed by positive regions, and decreases the uncertainty coming from boundary region.

关 键 词：单调性 下近似保持 下近似单调 变精度粗糙集 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]