Data-driven interactive multiobjective optimization using cluster based surrogate in discrete decision space
Abstract
Tutkielma esittää klusteripohjaisen sijaismallin diskreetin päätöksentekoavaruuden dimension pienentämiseksi ja lineaaristen kokonaislukuoptimointitehtävien yksinkertaistamiseksi. Sijaismalli on suunnattu erityisesti datapohjaisten päätöksenteko-ongelmien interaktiiviseen ratkaisemiseen, sillä se yhdistää sijaismallin interaktiota
helpottavan vaikutuksen ja interaktiivisen NIMBUS menetelmän hyvän
suorituskyvyn sijaismallin tuloavaruudessa. Kehitettyä sijaismallia
ja metodia myös sovellettiin monitavoitteiseen metsätalousongelmaan
hyvin tuloksin.
This thesis presents a cluster based surrogate model approach for reducing dimension of discrete decision space and so for simplifying integer linear optimization problems. The model is especially aimed for solving data-driven decision making problems interactively, as the surrogate makes interaction more seamless and the interactive NIMBUS method manages well within the product space of the surrogate. The developed cluster based surrogate and method were also applied for a Boreal Forest management problem with promising results.
This thesis presents a cluster based surrogate model approach for reducing dimension of discrete decision space and so for simplifying integer linear optimization problems. The model is especially aimed for solving data-driven decision making problems interactively, as the surrogate makes interaction more seamless and the interactive NIMBUS method manages well within the product space of the surrogate. The developed cluster based surrogate and method were also applied for a Boreal Forest management problem with promising results.
Main Author
Format
Theses
Master thesis
Published
2018
Subjects
The permanent address of the publication
https://urn.fi/URN:NBN:fi:jyu-201806133178Use this for linking
Language
English