DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Finding Mount Everest and Handling Voids

Storch, Tobias (2011) Finding Mount Everest and Handling Voids. Evolutionary Computation, 19 (2), pp. 325-344. MIT Press. doi: 10.1162/EVCO_a_00032.

Full text not available from this repository.


Evolutionary algorithms (EAs) are randomized search heuristics that solve problems successfully in many cases. Their behavior is often described in terms of strategies to find a high location on Earth’s surface. Unfortunately, many digital elevation models describing it contain void elements. These are elements not assigned an elevation. Therefore, we design and analyze simple EAs with different strategies to handle such partially defined functions. They are experimentally investigated on a dataset describing the elevation of Earth’s surface. The largest value found by an EA within a certain runtime is measured, and the median over a few runs is computed and compared for the different EAs. For the dataset, the distribution of void elements seems to be neither random nor adversarial. They are so-called semirandomly distributed. To deepen our understanding on the behavior of the different EAs, they are theoretically considered on well-known pseudo-Boolean functions transferred to partially defined ones. These modifications are also performed in a semirandom way. The typical runtime until an optimum is found by an EA is analyzed, namely bounded from above and below, and compared for the different EAs. We figure out that for the random model it is a good strategy to assume that a void element has a worse function value than all previous elements. Whereas for the adversary model it is a good strategy to assume that a void element has the best function value of all previous elements.

Item URL in elib:https://elib.dlr.de/69726/
Document Type:Article
Title:Finding Mount Everest and Handling Voids
AuthorsInstitution or Email of AuthorsAuthor's ORCID iD
Storch, Tobiastobias.storch (at) dlr.deUNSPECIFIED
Date:May 2011
Journal or Publication Title:Evolutionary Computation
Refereed publication:Yes
Open Access:No
Gold Open Access:No
In ISI Web of Science:Yes
DOI :10.1162/EVCO_a_00032
Page Range:pp. 325-344
Publisher:MIT Press
Keywords:Evolutionary algorithm, partially defined function, runtime analysis
HGF - Research field:Aeronautics, Space and Transport (old)
HGF - Program:Space (old)
HGF - Program Themes:W - no assignment
DLR - Research area:Space
DLR - Program:W - no assignment
DLR - Research theme (Project):W - no assignment (old)
Location: Oberpfaffenhofen
Institutes and Institutions:Remote Sensing Technology Institute > Photogrammetry and Image Analysis
Deposited By: Storch, Dr.rer.nat. Tobias
Deposited On:06 May 2011 07:41
Last Modified:06 Sep 2019 15:20

Repository Staff Only: item control page

Help & Contact
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.