A Preference-Based Interactive Evolutionary Algorithm for Multiobjective Optimization
2007 (English)Report (Other academic)
In this paper, we discuss ideas of incorporating preference information intoevolutionary multiobjective optimization and propose a preference-based evolutionaryalgorithm that can be used as an integral part of an interactive algorithm that we alsointroduce. At each iteration of the interactive algorithm, the DM is asked to givepreference information in terms of his/her reference point consisting of desirableaspiration levels for objective functions. The information is used in an evolutionaryalgorithm that generates a new population by combining the fitness function with anachievement scalarizing function containing the reference point. In the field ofmultiple criteria decision making, achievement scalarizing functions are widely usedfor projecting the reference point into the Pareto optimal set. In our approach, the nextpopulation is more concentrated in the area where more preferred alternatives areassumed to lie and the whole Pareto optimal set does not have to be generated withequal accuracy. The approach is demonstrated by numerical examples.
Place, publisher, year, edition, pages
Helsinki School of Economics Print, 2007.
, Working Papers, W-412
Multiple objectives, multiple criteria decision making, preference information, reference point, achievement scalarizing function
Economics and Business Computer and Information Science
IdentifiersURN: urn:nbn:se:kth:diva-83635ISBN: 978-952-488-094-7OAI: oai:DiVA.org:kth-83635DiVA: diva2:498868
QC 201202292012-02-122012-02-122012-02-29Bibliographically approved