Extreme Value Based Adaptive Operator Selection

Álvaro Fialho 1 Luis da Costa 2, 3 Marc Schoenauer 1, 2, 3, * Michèle Sebag 1, 2, 3
* Corresponding author
2 TAO - Machine Learning and Optimisation
LRI - Laboratoire de Recherche en Informatique, UP11 - Université Paris-Sud - Paris 11, Inria Saclay - Ile de France, CNRS - Centre National de la Recherche Scientifique : UMR8623
Abstract : Credit Assignment is a crucial ingredient for successful Adaptive Operator Selection. Instead of the average fitness improvement of newborn offspring, this paper proposes to use some empirical order statistics of those improvements, arguing that rare but highly beneficial jumps matter as much or more than frequent but small improvements. An extreme value based Credit Assignment is thus proposed, rewarding each operator with the best fitness improvement observed in a sliding window for this operator. This mechanism, combined with existing Adaptive Operator Selection rules, is investigated in an EC-like setting. First results show that the proposed method allows both the Adaptive Pursuit and the Dynamic Multi-Armed Bandit selection rules to actually track the best operators along evolution.
Document type :
Conference papers
Complete list of metadatas

Contributor : Álvaro Fialho <>
Submitted on : Wednesday, August 27, 2008 - 5:28:31 PM
Last modification on : Monday, December 9, 2019 - 5:24:06 PM
Long-term archiving on: Friday, May 28, 2010 - 7:51:06 PM


Files produced by the author(s)



Álvaro Fialho, Luis da Costa, Marc Schoenauer, Michèle Sebag. Extreme Value Based Adaptive Operator Selection. 10th International Conference on Parallel Problem Solving From Nature (PPSN X), Sep 2008, Dortmund, Germany. pp.175-184, ⟨10.1007/978-3-540-87700-4_18⟩. ⟨inria-00287355v1⟩



Record views


Files downloads