HYBRIDIZATION OF MAGNETIC CHARGE SYSTEM SEARCH METHOD FOR EFFICIENT DATA CLUSTERING

Abstract

MCSS is a relatively new meta-heuristic algorithm inspired from the electromagnetic theory and has shown better potential than the same class of algorithms. But, like the other meta-heuristic algorithm, some performance issues are also associated with this algorithm such as convergence rate and trap in local optima. So, in this work, an attempt is made to improve the convergence rate of MCSS algorithm and proposed a Hybrid Magnetic Charge System Search (HMCSS) for solving the clustering problems. Further, a local search strategy is also inculcated into MCSS algorithm to reduce the probability of trapping in local optima and exploring promise solutions. The effectiveness of the proposed algorithm is tested on some benchmark functions and also applied to solve real world clustering problems. The experimental results show that the proposed algorithm gives better results than the existing algorithms, and also improves the convergence rate of MCSS algorithm.