Reference sharing: a new collaboration model for cooperative coevolution
2017 (English)In: Journal of Heuristics, ISSN 1381-1231, E-ISSN 1572-9397, Vol. 23, no 1, 1-30 p.Article in journal (Refereed) Published
Cooperative coevolutionary algorithms have been a popular and effective learning approach to solve optimization problems through problem decomposition. However, their performance is highly sensitive to the degree of problem separability. Different collaboration mechanisms usually have to be chosen for particular problems. In the paper, we aim to design a collaboration model that can be successfully applied to a wide range of problems. We present a novel collaboration mechanism that offers this type of potential, along with a new sorting strategy for individuals that are assigned multiple fitness values. Furthermore, we demonstrate and analyze our algorithm through comparison studies with other popular cooperative coevolutionary models on a suite of standard function optimization problems.
Place, publisher, year, edition, pages
Springer, 2017. Vol. 23, no 1, 1-30 p.
Cooperative coevolution, Collaboration method, Fitness measurement, Function optimization
IdentifiersURN: urn:nbn:se:oru:diva-56841DOI: 10.1007/s10732-016-9322-9ISI: 000394215200001ScopusID: 2-s2.0-85010808949OAI: oai:DiVA.org:oru-56841DiVA: diva2:1085139