Ensemble clustering based on weighted co-association matrices: Error bound and convergence properties

详细信息    查看全文

• 作者：Vladimir Berikov^a ; ^b ; ^{berikov@math.nsc.ru}Author Vitae ; Igor Pestunov^b ; ^c ; ^{pestunov@ict.sbras.ru}Author Vitae
• 关键词：Weighted clustering ensemble ; Co-association matrix ; Latent variable model ; Cluster validity index ; Ensemble size ; Error bound ; Hyperspectral image segmentation
• 刊名：Pattern Recognition
• 出版年：2017
• 出版时间：March 2017
• 年：2017
• 卷：63
• 期：Complete
• 页码：427-436
• 全文大小：986 K
• 卷排序：63

文摘

An approach to ensemble clustering based on weighted co-association matrices is theoretically substantiated. The upper bound for misclassification probability in attributing a pair of observations to clusters is found. It is proved that clustering quality is improved with an increase in ensemble size and the expected evaluation function. Analytical dependencies between ensemble size and the quality estimates are derived.