Comparative Study of Recent Multimodal Evolutionary Algorithms

Abstract : Multimodal Optimization (MMO) aims at identifying several best solutions to a problem whereas classical optimization converge oftenly to only one good solution. MMO has been an active research area in the past years and several new evolutionary algorithms have been developed to tackle multimodal problems. In this work, we compare extensively three recent evolutionary algorithms (MoBiDE, Multimodal NSGAII and MOMMOP). Each algorithm uses multiobjectivization, together with niching techniques to address scalar (single objective) MMO problems. We have fully re-implemented MoBiDE and MM-NSGAII in order to better evaluate their sensitivity to parameter changes and their strengths and weaknesses. We have carefully evaluated all algorithms on the same benchmark functions and with the same parameters settings. The influence of the intrinsic parameters for each algorithm are stressed and the algorithms are also compared to a non-multimodal evolutionary algorithm to better highlight the impact of the multimodal adaptations. Moreover, full access to the detailed results and source code is granted on our website for the ease of reproducibility.
Type de document :
Communication dans un congrès
Computational Intelligence, 2015 IEEE Symposium Series on, 2015, Cape Town, South Africa. pp.837-844, 2015, Multicriteria Decision-Making. <10.1109/SSCI.2015.124>
Liste complète des métadonnées

https://hal-unice.archives-ouvertes.fr/hal-01322764
Contributeur : Denis Pallez <>
Soumis le : vendredi 27 mai 2016 - 16:17:15
Dernière modification le : samedi 28 mai 2016 - 01:13:45

Identifiants

Collections

Citation

R. Pighetti, Denis Pallez, Frédéric Precioso. Comparative Study of Recent Multimodal Evolutionary Algorithms. Computational Intelligence, 2015 IEEE Symposium Series on, 2015, Cape Town, South Africa. pp.837-844, 2015, Multicriteria Decision-Making. <10.1109/SSCI.2015.124>. <hal-01322764>

Partager

Métriques

Consultations de la notice

25