Adaptive geometric angle-based algorithm for pruning pareto-optimal sets

Conference proceedings article

Authors/Editors

Strategic Research Themes

No matching items found.

Publication Details

Author list: Sudeng S., Wattanapongsakorn N.

Publisher: Brno University of Technology

Publication year: 2013

Start page: 75

End page: 80

Number of pages: 6

ISBN: 9788021447554

ISSN: 1803-3814

eISSN: 1803-3814

URL: https://www.scopus.com/inward/record.uri?eid=2-s2.0-84905732532&partnerID=40&md5=c68464f8ff303faa66b8f22c663c3e81

Languages: English-Great Britain (EN-GB)

Abstract

Most of multi-objective evolutionary algorithms (MOEAs) approximate Pareto-optimal solutions covering wide area of the whole Pareto front favoring diversity property. The decision maker (DM) still works hard to compromise the large trade-off solutions. In this paper, we propose an algorithm to help the DM choosing the final best solution based on his/her preference. The main contribution of our algorithm is to filter out undesired solutions and provide more robust trade-off set of optimal solutions to the DM. Our algorithm is called an adaptive angle based pruning algorithm (ADA). The pruning rationale is to expand the dominated area for the purpose of removing solutions that only marginally improves in some objectives while being significantly worse in other objectives. Our pruning method begins by calculating the angle between a pair of solutions by using arctangent function. The bias intensity parameter is introduced as a minimum threshold angle in order to approximate the portions of desirable solutions based on the DM's preference. Initially, we consider several benchmark problems by applying a simple version of MOEA/D algorithm. Then, we apply the pruning algorithm. The experimental results have shown that our pruning algorithm provides robust sub-set of Pareto-optimal solutions for the benchmark problems. The pruned Pareto-optimal solutions distribute and cover multiple regions of Pareto front even when the strongest bias is applied. In addition, it is clearly shown in bi-objective problems that the pruned Pareto-optimal solutions are located at knee regions of the Pareto front.

Keywords

Adaptive angle-based pruning algorithm, Decision making analysis, Pareto-optimal solutions, Preference articulation techniques