摘要
信息系统属性的约简可以提高知识发现、机器学习等的精度和效率。本文提出了一种近似约简算法,该算法可使信息系统在基本保持原风格的情况下尽可能少地保留属性,为后期的系统处理节约了大量的处理时间。该算法的时间复杂度没有提高,约简后的属性大大减少。虽然原信息系统有一定的损失,但在一定的显著水平下是可以接受的。最后对一个有9个属性的信息系统进行了约简和近似约简的对比分析。
The attributes reduction of information system can enhance accuracy and efficiency of knowledge discovery, machine learning, etc. This paper proposes a approximately reduction arithmetic, this arithmetic can retain the minimal attributes in the basic form of the information systems, as much as possible to reduce attributes, that can save the time of the system's upper disposal. The arithmetic time order of the complexity has not enhanced, but the reduced attributes greatly reduced. The original information system has the certain loss, but this is may accept under the certain remarkable level. And a system having nine attributes been carried on the reduction and the approximately reduction of contrast analysis.
出处
《计算机科学》
CSCD
北大核心
2007年第7期165-167,共3页
Computer Science
基金
国家自然科学基金资助项目(批准号:60573056)
浙江省自然科学基金资助项目(批准号:Y105090)
关键词
属性约简策略
区分矩阵
近似约简
Attributes reduction strategy, Discernibility matrix, Approximately reduction