摘要
在已有的粗糙集属性约简算法基础上,给出了一个新的度量属性重要性的不可区分度函数,分析了不可区分度的性质,提出了一种能有效处理噪声的基于不可区分度的快速完备约简算法,最坏时间复杂度为max(O(|A||U|),O(|A|2|U/A|))。理论分析和实验结果表明,该约简算法在效率上较现有算法有显著提高,能较好抵制数据噪声,适于对大数据集进行处理。
After analyzing the attribute reduction algorithm based on rough set,a new definition of indiscernibility degree was given for rneasureing the importance of attribution, and the property of indiscernibility degree was analyzed. Then based on the indiscemibility degree, a new heuristic reduced algorithm was proposed, which is useful to deal with cut down to max（O（｜A｜｜U｜）,O（｜A｜^2｜U/A｜））. The theoretical analysis the noise and makes the worst time complexity and experimental results show that this new method is not only useful in solving data noise but also robust and efficient.
出处
《计算机科学》
CSCD
北大核心
2009年第8期196-200,共5页
Computer Science
基金
国家自然科学基金项目(60703013)资助
关键词
粗糙集
完备
约简
不可区分关系
Rough set, Complete, Reduction, Indiscernitility relation
