A performance assessment of dynamic heuristic optimisers on the IEEE CEC 2015 niching competition test problems

Abstract

We present the performance results of ve multi-modal heuristic optimisers on the 20 benchmark functions of the IEEE CEC 2015 competition on niching methods for multimodal optimisation. All the algorithms compared here are pre-de ned in existing works, are dynamic in their niche maintenance, and exploit `hill-valley' niche detection. Code and les of output statistics required by the competition are also provided online. We nd that although the recently developed Niching Migratory Multi-Swarm Optimiser (NMMSO) algorithm performs best overall, other optimisers perform better on some of the problems, with the main divergence in performance apparent between homogeneous and heterogeneous problem landscapes.