An Evolutionary Algorithm using GP surrogate model for expensive constrained optimization problems

DOI

10.2991/icista.2013.27How to use a DOI?

Keywords

Gaussian stochastic process model; Expensive constrained optimization;Surrogate model; Fitness evaluation; DyHF

Abstract

In expensive constrained optimization problems, the evaluation of candidate solutions could be extremely computationally and/or financially expensive. This paper proposes a method, called DyHF-GP, for reducing computation costs and raising optimization efficiency, by combining Gaussian stochastic process model(GP) with DyHF(Dynamic Hybrid Framework). In DyHF-GP, the Latin Hypercube Sampling(LHS) is used to sample points, then the true function is surrogated by GP. In evolutionary processes, the sample points and the GP are updated by retention and replacement mechanism. The using of GP and true function is controlled by error among several neighbor generations. The 13 standard test functions show that DyHF-GP has higher accuracy and retrieval efficiency. The number of FES is reduced by about 60% on average within 10-4 error, which diminishing the computation costs of the objective functions greatly.

Copyright

© 2013, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Meiyi Li
AU  - Hai Zhang
AU  - Rong Lv
PY  - 2013/06
DA  - 2013/06
TI  - An Evolutionary Algorithm using GP surrogate model for expensive constrained optimization problems
BT  - Proceedings of the 2013 International Conference on Information Science and Technology Applications (ICISTA-2013)
PB  - Atlantis Press
SP  - 133
EP  - 137
SN  - 1951-6851
UR  - https://doi.org/10.2991/icista.2013.27
DO  - 10.2991/icista.2013.27
ID  - Li2013/06
ER  -