Clustering performance comparison of new generation meta-heuristic algorithms

摘要

This article addressed two new generation meta-heuristic algorithms that are introduced to the literature recently. These algorithms, proved their performance by benchmark standard test functions, are implemented to solve clustering problems. One of these algorithms called Ions Motion Optimization and it is established from the motions of ions in nature. The other algorithm is Weighted Superposition Attraction and it is predicated on two fundamental principles, which are “attracted movements of agents” and “superposition”. Both of the algorithms are applied to different benchmark data sets consisted of continuous, categorical and mixed variables, and their performances are compared to Particle Swarm Optimization and Artificial Bee Colony algorithms. To eliminate the infeasible solutions, Deb's rule is integrated into the algorithms. The comparison results indicated that both of the algorithms, Ions Motion Optimization and Weighted Superposition Attraction, are competitive solution approaches for clustering problems.