Model Guided Sampling Optimization for Low-Dimensional Problems

0439763 - ÚI 2015 RIV PT eng C - Conference Paper (international conference)
Bajer, Lukáš - Holeňa, Martin
Model Guided Sampling Optimization for Low-Dimensional Problems.
ICAART 2015. Proceedings of the International Conference on Agents and Artificial Intelligence, Volume 2. Lisbon: Scitepress, 2015 - (Loiseau, S.; Filipe, J.; Duval, J.; van den Herik, J.), s. 451-456. ISBN 978-989-758-074-1.
[ICAART 2015. International Conference on Agents and Artificial Intelligence /7./. Lisbon (PT), 10.01.2015-12.01.2015]
R&D Projects: GA ČR GAP202/10/1333; GA ČR GA13-17187S
Institutional support: RVO:67985807
Keywords : black-box Optimization * Gaussian Process * Surrogate Modelling * EGO
Subject RIV: IN - Informatics, Computer Science

Optimization of very expensive black-box functions requires utilization of maximum information gathered by the process of optimization. Model Guided Sampling Optimization (MGSO) forms a more robust alternative to Jones’ Gaussian-process-based EGO algorithm. Instead of EGO’s maximizing expected improvement, the MGSO uses sampling the probability of improvement which is shown to be helpful against trapping in local minima. Further, the MGSO can reach close-to-optimum solutions faster than standard optimization algorithms on low dimensional or smooth problems.
Permanent Link: http://hdl.handle.net/11104/0242987