Effectiveness of Multi-step Crossover in Extrapolation Domain for Genetic Programming

Created by W.Langdon from gp-bibliography.bib Revision:1.8010

@InProceedings{Kuroda:2018:ieeeSMC,

• author = "Mao Kuroda and Yoshiko Hanada and Keiko Ono",
• booktitle = "2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC)",
• title = "Effectiveness of Multi-step Crossover in Extrapolation Domain for Genetic Programming",
• year = "2018",
• pages = "3654--3659",
• abstract = "In genetic programming (GP) which treats tree structured problems, inheritance and acquisition of traits in reproduction operators is one of most important issues. Two classes of search, focusing on inheritance or acquisition of traits, are called the interpolation search and the extrapolation search, respectively, by introducing a distance of solutions. Deterministic multi-step crossover fusion (dMSXF) is one of promising crossover that exploit parents' preferable trait, and it has been shown to perform very well on several tree structured problems. dMSXF is completely a interpolation-directed search, so that an exploration mechanism is required to explore traits that are not observed in parents. In several combinatorial optimization problems such as Travelling Salesman Problem, the effectiveness of incorporation of dMSMF into dMSXF has been shown, however its effectiveness has not been evaluated in tree optimization problems treated by GP. In this paper, we newly introduce deterministic multi-step mutation fusion (dMSMF), which is one of extrapolation-directed searches, to GP. Through the experiments using symbolic regression problems, we show the effectiveness of incorporation of the extrapolation-directed search to the interpolation-directed search in tree structured problems.",
• keywords = "genetic algorithms, genetic programming, Interpolation, Extrapolation, Optimisation, Search problems, Traveling salesman problems, Regression tree analysis",
• DOI = "doi:10.1109/SMC.2018.00618",
• ISSN = "2577-1655",
• month = oct,
• notes = "Also known as \cite{8616615}",

}

Genetic Programming entries for Mao Kuroda Yoshiko Hanada Keiko Ono

Citations