The Learning Power of Evolution
Abstract
We study the computational theory of evolution, asking whether Darwinian evolution can be modeled as a form of computational learning. We review the evolvability framework introduced by Valiant, which formalizes conditions under which a function class can be evolved by random variation and selection. We pose open problems about which function classes are efficiently evolvable and how evolvability relates to PAC learnability.
Cite
Text
Feldman and Valiant. "The Learning Power of Evolution." Annual Conference on Computational Learning Theory, 2008.Markdown
[Feldman and Valiant. "The Learning Power of Evolution." Annual Conference on Computational Learning Theory, 2008.](https://mlanthology.org/colt/2008/feldman2008colt-learning/)BibTeX
@inproceedings{feldman2008colt-learning,
title = {{The Learning Power of Evolution}},
author = {Feldman, Vitaly and Valiant, Leslie G.},
booktitle = {Annual Conference on Computational Learning Theory},
year = {2008},
pages = {513-514},
url = {https://mlanthology.org/colt/2008/feldman2008colt-learning/}
}