A Survey on Recent Multi-Objective Evolutionary Clustering Methods on Large Datasets
Clustering is widely used unsupervised learning method which provides effective solutions to the problems such as pattern recognition, market research and so on. Based on similarity, clustering technique creates homogenous groups. Clustering algorithms work on a single objective function wit prior knowledge on cluster structure. Such algorithms have limitations in many real life applications where multiple objective functions are to be considered. To overcome this problem, many algorithms came into existence with the concept of dealing with multiple objective functions in parallel. While performing clustering on real world data of an enterprise, it is essential to focus on multiple objectives rather than single objective. It is required in many domains including healthcare domain where it is essential to discover relationships among attributes with many objectives using clustering based solutions. This paper covers review of literature pertaining to multi-objective clustering in both discrete and continuous space. It provides insights on recent developments in multi-objective clustering research. It throws light into the evolutionary approaches considered, their merits and demerits. It also provides details pertaining to the multi-objective clustering algorithms in terms of their internal structures, optimisation strategies besides the effectiveness they achieved.