Projekte pro Jahr
Abstract
In simulation-based optimization, a common issue with many meta-heuristic algorithms is the limited computational budget. Performing a simulation is usually considerably more time-consuming than evaluating a closed mathematical function. Surrogate-assisted algorithms alleviate this problem by using representative models of the simulation which can be evaluated much faster. One of the most promising surrogate-assisted optimization approaches is Efficient Global Optimization, which uses Gaussian processes as surrogate-models. In this paper, the importance of carefully chosen hyper-parameters for Gaussian process kernels and a way of self-configuration is shown. Based on properties of the training set, e.g. distances between observed points, observed target values, etc., the hyper-parameters of the used kernels are initialized and bounded accordingly. With these initial values and bounds in mind, hyper-parameters are then optimized, which results in improved Gaussian process models that can be used for regression. The goal is to provide an automated way of hyper-parameter initialization, which can be used when building Kriging models in surrogate-assisted algorithms, e.g. Efficient Global Optimization (EGO). Obtained results show that applying the proposed hyper-parameter initialization and bounding can increase the performance of EGO in terms of either convergence speed or achieved objective function value.
Originalsprache | Englisch |
---|---|
Titel | 19th International Conference on Modeling and Applied Simulation, MAS 2020 |
Redakteure/-innen | Agostino G. Bruzzone, Fabio De Felice, Marina Massei, Adriano Solis |
Herausgeber (Verlag) | DIME UNIVERSITY OF GENOA |
Seiten | 60-67 |
Seitenumfang | 8 |
ISBN (elektronisch) | 9788885741492 |
DOIs | |
Publikationsstatus | Veröffentlicht - 2020 |
Veranstaltung | 19th International Conference on Modeling and Applied Simulation, MAS 2020 - Virtual, Online Dauer: 16 Sep. 2020 → 18 Sep. 2020 |
Publikationsreihe
Name | 19th International Conference on Modeling and Applied Simulation, MAS 2020 |
---|
Konferenz
Konferenz | 19th International Conference on Modeling and Applied Simulation, MAS 2020 |
---|---|
Ort | Virtual, Online |
Zeitraum | 16.09.2020 → 18.09.2020 |
Fingerprint
Untersuchen Sie die Forschungsthemen von „Hyper-parameter handling for Gaussian processes in efficient global optimization“. Zusammen bilden sie einen einzigartigen Fingerprint.Projekte
- 2 Abgeschlossen
-
JRZ adaptOp - Josef Ressel Zentrum für adaptive Optimierung in dynamischen Umgebungen
Wagner, S. (Leitende(r) Forscher/-in), Leitner, S. J. (Weitere Forschende), Beneder, M. (Weitere Forschende), Werth, B. (Weitere Forschende), Neuhauser, P. (Weitere Forschende), Heckmann, M. K. (Weitere Forschende), Karder, J. A. (Weitere Forschende), Beham, A. (Weitere Forschende) & Fleck, P. (Weitere Forschende)
01.10.2019 → 30.09.2024
Projekt: Forschungsprojekt
-
SimGenOpt2 - Integrated Methods for Robust Production Planning and Control
Beham, A. (Leitende(r) Forscher/-in), Altendorfer, K. (CoPI), Wagner, S. (Weitere Forschende), Affenzeller, M. (Weitere Forschende), Jodlbauer, H. (Weitere Forschende), Brunner, M. (Weitere Forschende), Werth, B. (Weitere Forschende), Seiringer, W. (Weitere Forschende) & Karder, J. A. (Weitere Forschende)
01.03.2017 → 01.03.2021
Projekt: Forschungsprojekt