基于max-⊙传递性的模糊聚类分析

Fuzzy Cluster Analysis Based on Max-⊙ Transitivity

由于∧算子太强 ,max -∧合成时丢失信息较多 ,聚类较粗糙 ,有时聚类结果很不理想 ,甚至无法聚类。研究了比它弱的⊙算子 ,在max -⊙传递性的基础上 ,提出了max -⊙合成和一种max -⊙聚类算法。因max -⊙合成法比较柔和 ,丢失信息较少 ,运用max -⊙聚类法进行聚类 ,效果更理想 ,它克服了max -min聚类法聚类粗糙的缺点。实例也证实了这一点。

Since ∧ operator is too strong ,it loses much information when max-∧ composition is performed, so cluster is rough. Sometimes, the result of partition is very bad, even the cluster can't be processed. The ⊙ operator is studied which is softer than ∧ operator, A max-⊙ composition and an algorithm based on the max-⊙ transitivity are proposed. Because the max-⊙ transitivity is softer, less information is lost when using it. So using max-⊙ cluster technique can obtain better results and overcome the disadvantage of max-min cluster method: roughness. The example verifies all these.