Comparison of different objects in multi-objective ensemble clustering

Abstract

Clustering is one of a greatest data mining tools that is used for partitioning dataset into different groups based on some similarity/dissimilarity metric. Traditional clustering algorithms often need prior knowledge about the data structure that makes clustering performance poorly when the cluster assumptions do not hold in the data sets. Multi objective clustering, in which multiple objective functions are simultaneously optimized, has emerged in such situations. In particular, application of multi objective evolutionary algorithms for clustering has become popular in the last decade because of their population-based nature. One of the most important case in multi objective evolutionary algorithms is objective functions that choose in evolutionary algorithms. In this paper we compare some different objects in evolutionary algorithm implemented with NSGA-II in ensemble clustering named MECA and compare the results between objectives.