Institutions | About Us | Help | Gaeilge
rian logo


Mark
Go Back
Managing Repetition in Grammar-Based Genetic Programming
Nicolau, Miguel; Fenton, Michael
Genetic and Evolutionary Computation - GECCO 2016, Genetic and Evolutionary Computation Conference, Denver, CO, USA, July 20-24, 2016, Proceedings, July, 2016 Grammar-based Genetic Programming systems are capable of generating identical phenotypic solutions, either by creating repeated genotypic representations, or from distinct genotypes, through their many-to-one mapping process. Furthermore, their initialisation process can generate a high number of duplicate individuals, while traditional variation and replacement operators can permit multiple individuals to percolate through generations unchanged. This can lead to a high number of phenotypically identical individuals within a population. This study investigates the frequency and effect of such duplicate individuals on a suite of benchmark problems. Both Grammatical Evolution and the CFG-GP systems are examined. Experimental evidence suggests that these useless evaluations can be instead be used either to speed-up the evolutionary process, or to delay convergence. Science Foundation Ireland
Keyword(s): Evolutionary algorithms; Genetic programming
Publication Date:
2017
Type: Other
Peer-Reviewed: Unknown
Language(s): English
Institution: University College Dublin
Publisher(s): ACM
First Indexed: 2017-01-05 05:25:14 Last Updated: 2018-10-11 15:05:20