Large scale global optimization: Experimental results with MOS-based hybrid algorithms

Abstract

Abstract-Continuous optimization is one of the most active research lines in evolutionary and metaheuristic algorithms. Through CEC 2005 to CEC 2013 competitions, many different algorithms have been proposed to solve continuous problems. The advances on this type of problems are of capital importance as many real-world problems from very different domains (biology, engineering, data mining, etc.) can be formulated as the optimization of a continuous function. In this paper we describe the whole process of creating a competitive hybrid algorithm, from the experimental design to the final statistical validation of the results. We prove that a good experimental design is able to find a combination of algorithms that outperforms any of its composing algorithms by automatically selecting the most appropriate heuristic for each function and search phase. We also show that the proposed algorithm obtains statistically better results than the reference algorithm DECC-G.

Constructive cooperative coevolution for large-scale global optimisation

“…These included the following algorithms: MOS (LaTorre et al 2013), IHDELS (Molina and Herrera 2015), CC-CMA-ES (Liu and Tang 2013), DECC-G (Yang et al 2008), VMODE (López et al 2015), MPS-CMA-ES (BolufeRohler et al 2015), jDEsps (Brest et al 2012), FBG-CMA-CC (Liu et al 2015), DECC-DG (Omidvar et al 2014). The results for these 9 algorithms were taken from literature (LaTorre et al 2015;López et al 2015;Bolufe-Rohler et al 2015;Liu et al 2015) The algorithms were ranked based on their reported mean performance for each one of the 15 benchmark functions in the CEC'2013 LSGO test suite and an overall ranking based on the average rank across the 15 functions was consequently calculated.…”

Designing benchmark problems for large-scale continuous optimization

“…Hybridization algorithms: MTSLS1-Reduced, Solis and Wets and GA minPart = 20%, stepFactor = 36,000 Refer to [28] for a detailed parameter setting of the constituent algorithms Table 6 Number of times each algorithm significantly outperforms other algorithms based on Table 5. Each row is based on a function category presented in Section 5.…”

“…Three major classes of algorithms are used for some preliminary comparative studies in this paper (see Section 7): traditional cooperative co-evolutionary, contribution-based cooperative co-evolution (CBCC) [41], and non-coevolutionary methods such as CMA-ES [20], JADE [69], SaDE [49], SaNSDE [66], Multiple Offspring Sampling (MOS) 1 [28] and standard differential evolution (DE) [56]. The contribution-based cooperative co-evolution allocates the available computational resources between subcomponents based on their contribution towards the improvement of the global objective value.…”

A novel hybridization of opposition-based learning and cooperative co-evolutionary for large-scale optimization

“…nWins is a N×N comparison method which compares each single method with all the others. Following [36], when an algorithm significantly outperforms one of its competitors, its nWins score is increased by +1 and the looser is penalized by −1. If both algorithms perform statistically similar, their nWins scores remain unchanged.…”