A Survey of Multiobjective Evolutionary Clustering

Abstract

Data clustering is a popular unsupervised data mining tool that is used for partitioning a given dataset into homogeneous groups based on some similarity/dissimilarity metric. Traditional clustering algorithms often make prior assumptions about the cluster structure and adopt a corresponding suitable objective function that is optimized either through classical techniques or metaheuristic approaches. These algorithms are known to perform poorly when the cluster assumptions do not hold in the data. Multiobjective clustering, in which multiple objective functions are simultaneously optimized, has emerged as an attractive and robust alternative in such situations. In particular, application of multiobjective evolutionary algorithms for clustering has become popular in the past decade because of their population-based nature. Here, we provide a comprehensive and critical survey of the multitude of multiobjective evolutionary clustering techniques existing in the literature. The techniques are classified according to the encoding strategies adopted, objective functions, evolutionary operators, strategy for maintaining nondominated solutions, and the method of selection of the final solution. The pros and cons of the different approaches are mentioned. Finally, we have discussed some real-life applications of multiobjective clustering in the domains of image segmentation, bioinformatics, web mining, and so forth.

Bio-inspired multiobjective clustering optimization: A survey and a proposal

“…The best solution is subjective and an expert can choose one or more solutions according to the problem requirements. [5,10,11] In a population-based algorithm a set of random points are started within the decision space, where each point represents a candidate solution for solving the multiobjective problem, the objective space has the mapped solution for the selected fitness functions. Figure 2a represents the solutions inside the decision space as circles and the ones in the objective space as squares and stars.…”

“…[20] They optimize two or more cluster validity indices (as well as the number of clusters) simultaneously, leading to high quality results, and have emerged as attractive and robust alternatives for clustering problems. [11] An important step to multiobjective clustering is the choice of suitable objective functions. Different cluster validity indices in different combinations are used in multiobjective clustering algorithms.…”

“…Different cluster validity indices in different combinations are used in multiobjective clustering algorithms. In Mukhopadhyay et al, [11] the authors divided the cluster validity indices into two groups: 1) Indices based on cluster prototypes; 2) Those based on cluster labels.…”

“…The most popular one is the multiobjective evolutionary clustering algorithm. [11] In Mukhopadhyay et al, [11] the authors present a review of multiobjective evolutionary clustering and introduce the main concepts of the area, such as the encoding strategies, choice of objective functions, effects of Published by Sciedu Press objective functions on the encoding strategies, main evolutionary operators and different approaches for generating the final solution. The authors also present an overview of different combinations of cluster validity indices normally used in multiobjective evolutionary clustering algorithms, and that can be extended to other non-evolutionary approaches, present in the literature, as summarized in Table 1.…”

“…[11] The genetic algorithms (GAs), firsly, were applied to solve mono objective optimization problems. The simple genetic algorithm (SGA) is used when an optimization problem has an unique objective.…”