一种基于加权特征的可能模糊聚类方法

利用数据点特征权重的概率约束关系和可能分布，提出了分别建立在概率和可能加权特征方式之上的改进可能模糊聚类的两种模型。其中建立在可能约束之上的改进PCM算法扩展了原算法，具有更广泛的适用性。实验结果表明，算法能够实现不同概率权重或可能分布特征条件下的模糊聚类，扩展了改进的PCM算法，适用性更广。与PCM及其改进算法相比，聚类的效果较为明显。

Possibility Fuzzy Clustering Approach with Weighted Features

Based on the different weights of features of objects, this paper presents two weighted feature-based improved possibilistic fuzzy clustering models separately with a probable weighted feature constraint and a possibilistic one respectively. The possibilistic model extends the PCM and enlarges its applications. Experimental results show that this model can reasonably cluster data in terms of different probable weights or possibilistic weights, which extends the applications of improved PCM, furthermore, the clustering results have an advantage over PCM.