On the performance of the hybridisation between migrating birds optimisation variants and differential evolution for large scale continuous problems

Abstract

Migrating Birds Optimisation (MBO) is a nature-inspired approach which has been shown to be very effective when solving a variety of combinatorial optimisation problems. More recently, an adaptation of the algorithm has been proposed that enables it to deal with continuous search spaces. We extend this work in two ways. Firstly, a novel leader replacement strategy is proposed to counter the slow convergence of the existing MBO algorithms due to low selection pressure. Secondly, MBO is hybridised with adaptive neighbourhood operators borrowed from Differential Evolution (DE) that promote exploration and exploitation. The new variants are tested on two sets of continuous large scale optimisation problems. Results show that MBO variants using adaptive, exploration-based operators outperform DE on the CEC benchmark suite with 1000 variables. Further experiments on a second suite of 19 problems show that MBO variants outperform DE on 90% of these test-cases.