基于粗糙集和感知器的模式分类方法

A Pattern Classification Method Based on Rough Set and Pereceptron

基于粗糙集相容关系给出了一个模式分类算法。先通过感知器神经网络训练属性相容权值和相容阈值,再由相容关系确定每个样本的上下近似,通过引入一个用于决策的粗糙隶属函数,给出了分类算法。最后,对IR IS数据集的实验结果表明本算法是有效的。

In this paper,a pattern classification based on tolerant rough set is given.First,tolerant weight and tolerant threshold are trained by pereceptron.Then,the low and upper approximation sets of training set are given using the tolent relation.A rough membership function to make a decision is introduced and a classify algorithm is given.Finally,we experiment on IRIS data.It is proved that this algorithm is effective.