An Efficient Hybrid Fuzzy Clustering Method

It is known that fuzzy algorithms will stop minimizing the objective function whenever reaches to a local minimum. It is known also that they are biased with the initial values of input parameters. In this work we address the minimizing problem of the objective function in the field of fuzzy clustering and introduce a method which is based on the majorization idea and the KNN algorithm. It speeds up the search for the minimum, passes the local minimum, and also tries to set the initial values into sound values. We use a modified version of the KNN in such a way that the initial values for the parameters can be settled. There are two approaches for dealing with local minima. The algorithm has been put into study by applying it into three known datasets. The results show that the performance of the proposed method is promising