基于变精度模型的平分互测动态约简粗糙集知识发现方法研究

Rough set knowledge acquisition of“divide equally and examine each other”and“dynamic reduce”based on variable precision model

普通粗糙集模型对数据噪音的高度敏感限制了其在工程实际中的应用,本文在变精度模型近似约简的基础上提出了数据全集随机平分互测法以提高数据的利用率。为克服数据集随机分割带来的约简值浮动变化的问题,本文提出了动态约简的方法筛选出最优约简,将此最优约简应用于数据全集生成最优规则。

Owing to the high sensitivity to noise data, the application of normal rough set model in engineering is restricted, this thesis put forward the method of “data set divided equally and examine each other”to improve data utilization ratie. In order to solve the problem of the fluctuation of reduction with the data set random division, this thesis bring forward the method of dynamic reduce to confirm the best reduction, and succeed in obtaining the optimum rule by applying the best reduction to data set.