Publication Details

Bayesovské optimalizační algoritmy v dynamickém prostředí

KOBLIHA Miloš. Bayesovské optimalizační algoritmy v dynamickém prostředí. In: Sborník příspevků ze semináře Počítačové Architektury & Diagnostika. Praha: Czech Technical University, 2005, p. 6. ISBN 80-01-03298-1.
English title
Bayesian Optimizing Algorithms in Dynamic Environment
Type
conference paper
Language
czech
Authors
Kobliha Miloš, Ing. (DCSY FIT BUT)
Keywords

BOA, dynamic problems, evolutionary optimization

Abstract

This paper is an experimental study investigating the capability of Bayesian optimization algorithms to solve dynamic problems. We tested the performance of two types of Bayesian optimization algorithms - Mixed continuous-discrete Bayesian Optimization Algorithm (MBOA) [1], and Adaptive Mixed Bayesian Optimization Algorithm (AMBOA)[2]. We have compared the behaviour of both algorithms on a simple dynamic environment defined as a time-varying function with predefined parameters. The experimental results confirmed the capability of both BOA algorithms to adapt the search process, but for a limited environment change. The AMBOA with the variance adaptation outperformed the MBOA algorithm.

Published
2005
Pages
6
Proceedings
Sborník příspevků ze semináře Počítačové Architektury & Diagnostika
Conference
Počítačové architektury a diagnostika - Pracovní seminář pro studenty doktorského studia, Lázně Sedmihorky, CZ
ISBN
80-01-03298-1
Publisher
Czech Technical University
Place
Praha, CZ
BibTeX 
@INPROCEEDINGS{FITPUB8084,
   author = "Milo\v{s} Kobliha",
   title = "Bayesovsk\'{e} optimaliza\v{c}n\'{i} algoritmy v dynamick\'{e}m prost\v{r}ed\'{i}",
   pages = 6,
   booktitle = "Sborn\'{i}k p\v{r}\'{i}spevk\r{u} ze semin\'{a}\v{r}e Po\v{c}\'{i}ta\v{c}ov\'{e} Architektury \& Diagnostika",
   year = 2005,
   location = "Praha, CZ",
   publisher = "Czech Technical University",
   ISBN = "80-01-03298-1",
   language = "czech",
   url = "https://www.fit.vut.cz/research/publication/8084"
}
Back to top