Neural Networks as Surrogate Models for Measurements in Optimization Algorithms

0345993 - ÚI 2011 RIV DE eng C - Conference Paper (international conference)
Holeňa, Martin - Linke, D. - Rodemerck, U. - Bajer, Lukáš
Neural Networks as Surrogate Models for Measurements in Optimization Algorithms.
Analytical and Stochastic Modeling Techniques and Applications. Berlin: Springer, 2010 - (Al-Begain, K.; Fiems, D.; Knottenbelt, W.), s. 351-366. Lecture Notes in Computer Science, 6148. ISBN 978-3-642-13567-5. ISSN 0302-9743.
[ASMTA 2010. International Conference /17./. Cardiff (GB), 14.06.2010-16.06.2010]
R&D Projects: GA ČR GA201/08/0802
Institutional research plan: CEZ:AV0Z10300504
Keywords : functions evaluated via measurements * evolutionary optimization * surrogate modelling * neural networks * boosting
Subject RIV: IN - Informatics, Computer Science

The paper deals with surrogate modelling, a modern approach to the optimization of objective functions evaluated via measurements. The approach leads to a substantial decrease of time and costs of evaluation of the objective function, a property that is particularly attractive in evolutionary optimization. The paper recalls common strategies for using surrogate models in evolutionary optimization, and proposes two extensions to those strategies - extension to boosted surrogate models and extension to using a set of models. These are currently being implemented, in connection with surrogate modelling based on feed-forward neural networks, in a software tool for problem-tailored evolutionary optimization of catalytic materials. The paper presents results of experimentally testing already implemented parts and comparing boosted surrogate models with models without boosting, which clearly confirms the usefulness of both proposed extensions.
Permanent Link: http://hdl.handle.net/11104/0187137