Publication Details
Bayesian Optimization Algorithms for Dynamic Problems
Schwarz Josef, doc. Ing., CSc. (DCSY FIT BUT)
Očenášek Jiří, Ing., Ph.D. (DCSY FIT BUT)
BOA algorithm, dynamic problem, optimalization
This paper is an experimental study investigating the capability of Bayesian optimization algorithms to solve dynamic problems. We tested the performance of two variants of Bayesian optimization algorithms - Mixed continuous-discrete Bayesian Optimization Algorithm (MBOA), Adaptive Mixed Bayesian Optimization Algorithm (AMBOA) - and new proposed modifications with embedded Sentinels concept and Hypervariance. We have compared the performance of these variants on a simple dynamic problem - a time-varying function with predefined parameters. The experimental results confirmed the benefit of Sentinels concept and Hypervariance embedded into MBOA algorithm for tracking a moving optimum.
@INPROCEEDINGS{FITPUB8083, author = "Milo\v{s} Kobliha and Josef Schwarz and Ji\v{r}\'{i} O\v{c}en\'{a}\v{s}ek", title = "Bayesian Optimization Algorithms for Dynamic Problems", pages = "800--804", booktitle = "Applications of Evolutionary Computing", journal = "Lecture Notes in Computer Science", volume = 2006, number = 3907, year = 2006, location = "Budapest, HU", publisher = "Springer Verlag", ISBN = "3-540-33237-5", ISSN = "0302-9743", language = "english", url = "https://www.fit.vut.cz/research/publication/8083" }