New possibilistic noise rejection clustering algorithm with simulated annealing

Fuzzy C-Means has been used as a popular fuzzy clustering method due to its simplicity and high speed in clustering large data sets. However, C-Means has two shortcomings: dependency on the initial state and convergence to local optima. In this paper a new algorithm based on simulated annealing and possibilistic noise rejection clustering is proposed to reduce the problem of converging to local minima and dependency on initial states. The comparison of the proposed algorithms and some other algorithms in the literature shows that the algorithms outperforms other algorithms in terms of optimization objective function and is capable of doing clustering in noisy environments more efficiently.