鲁棒区间类型2可能性C均值聚类

Robust interval type-2 possibilistic C-means clustering

提出了一种鲁棒区间类型2可能性C均值(IT2PCM)聚类规则,其实质是采用交替迭代结构进行聚类的交替类估计,但隶属度函数通过区间类型2模糊集合来选择.在提出的方法中,类原型的更新方程通过将降型与解模糊相结合的形式来计算.在鲁棒统计的框架下,通过φ函数的分析指出这种更新方程对类内不确定的模式以及野点具有鲁棒性.最后,与现有的鲁棒聚类规则进行比较,仿真结果说明了IT2PCM良好的鲁棒性.

This paper presents alternating iteration architecture for clustering, robust interval type-2 possibilistic C-means （IT2PCM） clustering algorithm. It is actually alternating cluster estimation, but membership functions are selected directly with interval type-2 fuzzy sets by the users. In the proposed method, the cluster prototype update equation is calculated by type reduction combined with defuzzification. It is robust to uncertain inliers and outtiers on the basis of its φ function analysis in the framework of robust statistics. Simulation results of comparing IT2PCM with existing methods show the nice robust properties of IT2PCM.