A surrogate-assisted selection scheme for genetic algorithms employing multi-layer neural networks

Masaki Fujiwara, Masaharu Munetomo

Abstract

In this paper, we propose a simple yet effective approach in surrogate-assisted genetic algorithms employing a neural network to estimate survival probabilities of individuals in selections to reduce computational cost of their fitness evaluations. A surrogate-assisted selection scheme we propose employs multi-layer perceptron to learn whether an individual is selected or not without fitness evaluations based on the observations of selections by fitness evaluations in the previous generations.
BibTeX
@inproceedings{fujiwara2018surrogate,
  author = {Masaki Fujiwara and Masaharu Munetomo},
  title = {{A surrogate-assisted selection scheme for genetic algorithms employing multi-layer neural networks}},
  booktitle = {GECCO (Companion)},
  year = {2018},
  doi = {10.1145/3205651.3208787},
}

Loading related content...

Figure