Going faster and hence further with lexicase selection

Li Ding, Ryan Boldi, Thomas Helmuth, Lee Spector

Abstract

Lexicase selection is a semantic-aware parent selection method, which has demonstrated success in multiple research areas including genetic programming, symbolic regression, and recently deep learning. One potential drawback of lexicase selection and its variants is that the selection procedure requires evaluating training cases in a single data stream, making it difficult to handle tasks where the evaluation is computationally heavy or the dataset is large-scale. In this work, we investigate how the weighted shuffle methods can be employed to improve the efficiency of lexicase selection. We propose fast lexicase selection, a method that incorporates lexicase selection and weighted shuffle with partial evaluation. Experiments on both classic genetic programming and deep learning tasks demonstrate that the proposed method has superior efficiency with a significantly reduced number of evaluation steps during selection, as well as sustains equivalent performance.
BibTeX
@inproceedings{ding2022going,
  author = {Li Ding and Ryan Boldi and Thomas Helmuth and Lee Spector},
  title = {{Going faster and hence further with lexicase selection}},
  booktitle = {GECCO Companion},
  year = {2022},
  doi = {10.1145/3520304.3529059},
}

Loading related content...

Figure