The art of parameterless evolutionary algorithms

"Evolutionary Algorithms (EAs) can be powerful problem solving tools when other methods fail. EAs maintain a collection of solutions and go through many iterations of recombining and randomly modifying current solutions, deleting some of the discovered solutions and giving the better solutions a higher chance of surviving. In such a manner EAs explore the search space in the pursuit of globally optimal solutions. To make an EA successful in its pursuit of globally optimal solutions on any particular problem, the evolutionary operators and relevant parameters must be carefully tuned. The tuning of operators and parameters is often manual and time consuming. The focus of this dissertation is on automatic tuning of operators and parameters. This dissertation presents methodologies for automating the tuning of the population size parameter and the choice of the mate selection operator"--Abstract, page iii.

Name thesis:degree_name

Ph. D. in Computer Science

Grantor

Missouri University of Science and Technology

Year dc:date.available

2016

Author dc:creator

• Holdener, Ekaterina A.

• Computer Sciences

Repository record dc:identifier

https://scholarsmine.mst.edu/doctoral_dissertations/1759

OAI identifier oai:identifier

oai:scholarsmine.mst.edu:doctoral_dissertations-2761