Promoting Speciation Through Variable Dominance in Genetic Algorithms

Simple item page
dc.contributor.advisorHougen, Dean
dc.contributor.authorDaro, Robert
dc.contributor.committeeMemberFagg, Andrew H.
dc.contributor.committeeMemberSchlupp, Ingo
dc.date.accessioned2020-05-19T18:29:12Z
dc.date.available2020-05-19T18:29:12Z
dc.date.issued2020-05
dc.date.manuscript2020-05-08
dc.description.abstractGenetic algorithms are a class of search algorithms that have been around since the 1970s. Despite their age, genetic algorithms still see a great deal of use in various applications and so many efforts have gone into addressing some of their limitations. Chief among the genetic algorithm's limitations is its tendency to converge to singular answers even when working with problem spaces that feature multiple optima. Furthermore, genetic algorithms will often converge to answers that are only local optima and not globally optimum. To that end, various researchers have developed methods to both encourage exploration of a problem space as well as promote diversification of genes to avoid global convergence. The most promising method is speciation, a term and concept borrowed from biology. New research shows that genetic structures using diploid chromosome models, meiotic inheritance, and gene dominance encourage speciation without the need for expensive selection modifications. Current implementations of dominance are hypothesized to skew exploration of search spaces. This work modifies the diploid approach to encourage exploration more generally by allowing dominance to evolve alongside the genes affecting the phenotype of the chromosome. These changes encourage speciation by allowing more symmetric exploration of the search space. While the modifications are not perfect, they serve as an excellent proof of concept for future work.en_US
dc.identifier.urihttps://hdl.handle.net/11244/324424
dc.languageen_USen_US
dc.rightsAttribution-ShareAlike 4.0 International*
dc.rights.urihttp://creativecommons.org/licenses/by-sa/4.0/*
dc.subjectGenetic Algorithmsen_US
dc.subjectSpeciationen_US
dc.subjectArtificial Intelligenceen_US
dc.subjectMachine Learningen_US
dc.thesis.degreeMaster of Scienceen_US
dc.titlePromoting Speciation Through Variable Dominance in Genetic Algorithmsen_US
ou.groupGallogly College of Engineering::School of Computer Scienceen_US

Files

Original bundle
Now showing 1 - 2 of 2
Thumbnail Image
Name:
2020_Daro_Robert_Thesis.pdf
Size:
142.97 MB
Format:
Adobe Portable Document Format
Description:
Download
No Thumbnail Available
Name:
2020_Daro_Robert_Thesis.tar
Size:
164.09 MB
Format:
Unknown data format
Description:
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

OU - Theses