ML Anthology

When Will a Genetic Algorithm Outperform Hill Climbing

Melanie Mitchell, John H. Holland, Stephanie Forrest
NeurIPS 1993 pp. 51-58
/neurips/1993/mitchell1993neurips-genetic/

Abstract

We analyze a simple hill-climbing algorithm (RMHC) that was pre(cid:173) viously shown to outperform a genetic algorithm (GA) on a simple "Royal Road" function. We then analyze an "idealized" genetic algorithm (IGA) that is significantly faster than RMHC and that gives a lower bound for GA speed. We identify the features of the IGA that give rise to this speedup, and discuss how these features can be incorporated into a real GA.

PDF NeurIPS Semantic Scholar

Cite

Text

Mitchell et al. "When Will a Genetic Algorithm Outperform Hill Climbing." Neural Information Processing Systems, 1993.

Markdown

[Mitchell et al. "When Will a Genetic Algorithm Outperform Hill Climbing." Neural Information Processing Systems, 1993.](https://mlanthology.org/neurips/1993/mitchell1993neurips-genetic/)

BibTeX

@inproceedings{mitchell1993neurips-genetic,
  title     = {{When Will a Genetic Algorithm Outperform Hill Climbing}},
  author    = {Mitchell, Melanie and Holland, John H. and Forrest, Stephanie},
  booktitle = {Neural Information Processing Systems},
  year      = {1993},
  pages     = {51-58},
  url       = {https://mlanthology.org/neurips/1993/mitchell1993neurips-genetic/}
}