摘要
针对变精度粗糙集理论的属性约简问题,从属性依赖度增量、互信息增量以及条件属性相对于决策属性的覆盖度与准确度相结合的量度等角度对属性重要度进行分析,并分别以这三个属性重要度的度量作为启发式信息,给出变精度粗糙集属性约简的启发式算法,同时将以上算法应用Matlab程序进行实现。最后,给出具体的算例利用所给的算法求得信息系统的在不同启发式信息下的属性约简,并对所得的结果进行比较,从而说明所给算法的有效性和实用性。
Heuristic algorithms are proposed to solve the attribute reduction of variable precision rough sets theory. Attribute significance is considered from the increment of dependence of attribute, the increment of mutual information and the measure of combine degree of nicety with degree of coverage, and which is regarded as heuristic information in the heuristic algorithms. These corresponding algorithms are realized by the program of Matlab. The validity and practicability of these algorithms is proven, and the different attribution reductions of the information system are shown using heuristic algorithms through a practical example.
出处
《控制工程》
CSCD
2007年第5期505-507,共3页
Control Engineering of China
基金
国家自然科学基金资助项目(70271066)
关键词
变精度粗糙集
β约简
准确度
覆盖度
variable precision rough sets
β reduction
degree of confidence
degree of coverage
