不完备决策系统中规则提取的快速矩阵算法  

Fast Matrix Computation Algorithm for Rules Extraction in Incomplete Decision Systems

作　　者：程玉胜[1] 张佑生[2] 胡学钢[2] 

机构地区：[1]安庆师范学院计算机与信息学院,安徽安庆246011 [2]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《系统仿真学报》2008年第15期4036-4040,共5页Journal of System Simulation

基　　金：安徽省自然科学基金(070412061)

摘　　要：提出了广义决策的不可分辨关系及其不可分辨类、联合决策相容矩阵等概念以及不完备决策系统中基于联合决策相容矩阵的约简和规则提取的快速矩阵算法。将条件属性相容矩阵和决策属性分配决策矩阵合并到一个矩阵中,大大减少了矩阵生成过程的比较次数,有效的提高了算法的效率。通过一个实例,说明了算法的执行过程;算法复杂度分析以及相应的对比实验,进一步说明了该方法的有效性。Some new concepts were defined such as indiscernibility relation of general decision, the indiscernibility classes, joint decision tolerance matrix and a fast computation algorithm was proposed for reduction and rule extraction based on joint decision matrix in incomplete decision systems. By putting the tolerance matrix of conditional attribute and assignment decision matrix of decision attribute into one matrix, the comparison times was considerably reduced in matrix forming process, which could considerably improve the algorithm efficiency. Through an example, the algorithm carrying-out was accounted for. The complexity analysis and a corresponding comparative experiment show that the method is quite effect.

关 键 词：粗糙集理论 相容矩阵 联合决策相容矩阵 不完备决策系统 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]