Improving data clustering using fuzzy logic and PSO algorithm

Intelligent algorithms have always been used as a global search method in many optimization problems. One of these problems is clustering problem. Clustering is a kind of process which receives a set of data as input and classifies them into several sub-groups. Clustering algorithms which use fuzzy measure, such as FCM, have obvious advantages over explicit samples. Despite advantages of FCM in group determination over similar explicit method, first the number of clusters and their centers should be determined optionally and there is a high probability for being trapped in local peaks. Therefore we present a new algorithm which avoids being trapped in local peaks which uses fuzzy logic and PSO algorithm and finds global optimal response or optimal place of cluster centers. All of results indicate the priority of the proposed algorithm.