A new GA based method for improving hybrid clustering

In this paper a new hybrid clustering method is presented which uses fuzzy logic and genetic algorithm. There are two main phases that should be investigated. The first is coding hybrid clustering problem in a way that could be solved by genetic algorithm. The other is designing an evaluation function which conducts the potential results to the global optimum. In this paper, a novel fuzzy criterion for evaluating the final partition is proposed which uses string representation of ensemble of primary clusters. The objective function tries to maximize the agreement between the ensemble members as well as minimize the disagreement simultaneously. The efficiency of the proposed method is evaluated by multiple standard databases. The promising obtained results show the outperforming of the proposed method compared to the other well known clustering method.