基于近似精度递归计算的一个属性约简算法

Attribute Reduction Algorithm Based on Recursive Computing of Approximation Quality

首先分析粗糙集中不同正区域之间的关系 ,在此基础上推导出近似精度计算的一个简洁的递归公式 .在Jelonek属性约简算法[3 ] 的基础上 ,应用上述递归公式和独立性条件判别策略设计了一个改进的属性约简算法 ,与原算法相比 ,不仅在算法速度提高方面取得了明显的效果 ,而且保证了约简算法的正确性 .

In this paper,we analyze firstly the relationship between positive regions in the context of rough set and in light of which we derive a simple recursive formula for computing the approximation quality.Then,the formula together with a strategy for verifying the independency condition is applied to improve Jelonek's attribute reduction algorithm.It turns out that the improved algorithm not only efficiently enhances the speed but also ensures the correctness of the algorithm.