基于可能性分布的聚类有效性

Cluster Validity Based On Possibiliatic Distribution

依据可能性理论，本文引入了一个新的划分系数。结合Ｊ．Ｃ．Ｂｅｚｄｅｋ给出的划分系数，定义一个新的聚类有效性函数。计算机模拟表明该聚类有效性函数具有良好判决功能和鲁棒性。

Based on the possibility theory,a new partition coefficient is introduced. By combinating thisnew partition coefficient with the partition coefficient defned by J . C . Bezdek , we give a new cluster validity func-tion. Experimental results show this cluster validity function's effectiveness and robustness.