A NEW APPROACH FOR DATA CLUSTERING USING MULTI-VERSE OPTIMIZER ALGORITHM

Abstract

Nature has always been a source of inspiration. Over the last few decades, it has stimulated many successful algorithms and computational tools for dealing with complex and optimization problems. This work proposes a new heuristic algorithm that is inspired by the Universe theory of multi-verse i.e. more than one Universe phenomenon. Similar to other population-based algorithms, the Multi-verse optimizer (MVO) starts with an initial population of candidate solutions to an optimization problem and an objective function that is calculated for them. At each iteration of the MVO, the best candidate is selected to be the Best Universe, which then starts exchanging the objects from other Universe. Also the Universes with high inflation rate move their objects to the universe having low inflation rate in order to make abrupt changes. To evaluate the performance of the MVO algorithm, it is applied to solve the clustering problem, which is a NP-hard problem. The experimental results show that the proposed MVO clustering algorithm outperforms other traditional heuristic algorithms for five benchmark datasets.