A Framework to Hybridize PBIL and a Hyper-heuristic for Dynamic Environments

Gönül Uluda¹, Berna Kiraz¹, A. ima Etaner-Uyar¹, and Ender Özcan²

¹Istanbul Technical University, Turkey
uludagg@itu.edu.tr
berna.kiraz@marmara.edu.tr
etaner@itu.edu.tr

²University of Nottingham, UK
Ender.Ozcan@nottingham.ac.uk

Abstract. Selection hyper-heuristic methodologies explore the space of heuristics which in turn explore the space of candidate solutions for solving hard computational problems. This study investigates the performance of approaches based on a framework that hybridizes selection hyper-heuristics and population based incremental learning (PBIL), mixing offline and online learning mechanisms for solving dynamic environment problems. The experimental results over well known benchmark instances show that the approach is generalized enough to provide a good average performance over different types of dynamic environments.

Keywords: hyper-heuristics, dynamic environments, multiple populations, incremental learning

LNCS 7492, p. 358 ff.

Full article in PDF | BibTeX

lncs@springer.com
© Springer-Verlag Berlin Heidelberg 2012