Making Hard(er) Benchmark Functions: Genetic Programming

Dante Niewenhuis, Abdellah Salhi, Daan van den Berg, Daan van den Berg

2024

Abstract

TreeEvolver, a genetic programming algorithm, is used to make continuous mathematical functions that give rise to 3D landscapes. These are then empirically tested for hardness by a simple evolutionary algorithm, after which TreeEvolver mutates the functions in an effort to increase the hardness of the corresponding landscapes. Results are wildly diverse, but show that traditional continuous benchmark functions such as Branin, Easom and Martin-Gaddy might not be hard at all, and much harder objective landscapes exist.

Paper Citation

in Harvard Style

Niewenhuis D., Salhi A. and van den Berg D. (2024). Making Hard(er) Benchmark Functions: Genetic Programming. In Proceedings of the 26th International Conference on Enterprise Information Systems - Volume 1: ICEIS; ISBN 978-989-758-692-7, SciTePress, pages 567-578. DOI: 10.5220/0012475800003690

in Bibtex Style

@conference{iceis24,
author={Dante Niewenhuis and Abdellah Salhi and Daan van den Berg},
title={Making Hard(er) Benchmark Functions: Genetic Programming},
booktitle={Proceedings of the 26th International Conference on Enterprise Information Systems - Volume 1: ICEIS},
year={2024},
pages={567-578},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012475800003690},
isbn={978-989-758-692-7},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 26th International Conference on Enterprise Information Systems - Volume 1: ICEIS
TI - Making Hard(er) Benchmark Functions: Genetic Programming
SN - 978-989-758-692-7
AU - Niewenhuis D.
AU - Salhi A.
AU - van den Berg D.
PY - 2024
SP - 567
EP - 578
DO - 10.5220/0012475800003690
PB - SciTePress