Beyond No Free Lunch: Realistic Algorithms for Arbitrary Problem Classes
Information Theory
2010-03-17 v3 Neural and Evolutionary Computing
math.IT
Abstract
We show how the necessary and sufficient conditions for the NFL to apply can be reduced to the single requirement of the set of objective functions under consideration being closed under permutation, and quantify the extent to which a set of objectives not closed under permutation can give rise to a performance difference between two algorithms. Then we provide a more refined definition of performance under which we show that revisiting algorithms are always trumped by enumerative ones.
Keywords
Cite
@article{arxiv.0907.1597,
title = {Beyond No Free Lunch: Realistic Algorithms for Arbitrary Problem Classes},
author = {James A. R. Marshall and Thomas G. Hinton},
journal= {arXiv preprint arXiv:0907.1597},
year = {2010}
}
Comments
(V3 fixed some other typos and improved some results)