Affiliations: [a] Department of Electronic Engineering, University of Electronic Science and Technology of China, Chengdu, China
| [b] Research Institute of Electronic Science and Technology, University of Electronic Science and Technology of China, Chengdu, China
Correspondence:
[*]
Corresponding author. Zhichao Sun, Department of Electronic Engineering, University of Electronic Science and
Technology of China, Chengdu 611731, China. Tel.: +86 13699441219; Fax: +86 028 61830054;
E-mail: [email protected].
Abstract: Possibilistic fuzzy c-means (PFCM) is used for solving the problem of data classification. It relies on initial cluster centers set by users that are lack of theoretical supporting. Inappropriate initial values may result in deviation of cluster centers. In this paper, A Robust Adaptive Particle Swarm Optimization based on steepest descent method is proposed to solve the problem of initialization and improving the performance of clustering. Combined with clustering algorithm, particle swarm optimization (PSO) possesses the good robustness to noises. Furthermore, since traditional PSO is inefficient when searching in the complex nonlinear hyperspace. Steepest descend method is applied to adaptively adjusting parameters. Moreover, optimum combined position is used to update the current information of each particle, which can discover more useful information lies in personal optimal experience and global optimal experience. The performance of proposed algorithm are tested in numerical simulations. The effectiveness, accuracy and stability of the new model are verified by simulations both with and without noises.
