Creating and sharing knowledge for telecommunications

Enhancing Evidence Accumulation Clustering using PCA

Duarte, J. ; Aidos, H. ; Fred, A. L. N.

Enhancing Evidence Accumulation Clustering using PCA, Proc Jornadas de Classificação e Análise de Dados - JOCLAD, Tomar, Portugal, Vol. 0, pp. 120 - 123, March, 2012.

Digital Object Identifier:

Abstract
The Evidence Accumulation Clustering (EAC) is a well-known clustering combination approach which combines multiple data partitions into a co-association matrix and then produces the consensus partition by applying a clustering algorithm to it. We propose an extension to the EAC method by cleaning the noise in the co-association matrix using an approach similar to the Principal Components Analysis. Experimental results on several synthetic and real data sets shown that the proposed approach improve the performance of EAC.