论文部分内容阅读
提出了一种标准粗糙集约简时连续属性离散化的新方法。采用标准粗糙集进行属性约简时,要求属性为离散的,而大多数情况下属性是连续的,因此需要进行离散化处理。首先介绍了原有的信息熵算法并指出其局限性;其次,对多类别信息熵进行扩充,将距离因素引入到该信息熵的计算中;最后给出了扩展信息熵计算的两个基本准则,利用证据理论完成信度的上聚焦。仿真显示了该方法的有效性。
A new method of discretization of continuous attributes is proposed when standard rough sets are reduced. Attribute reduction using standard rough sets requires that attributes be discrete, whereas attributes are continuous in most cases and therefore require discretization. Firstly, the original information entropy algorithm is introduced and its limitations are pointed out. Secondly, the multi-category information entropy is extended, and the distance factor is introduced into the calculation of this information entropy. Finally, two basic rules of extended information entropy computing are given. , The use of evidence theory to complete the reliability of the focus. Simulation shows the effectiveness of this method.