Free lunches for neural network search

Riccardo Poli, Mario Graff
2009 Proceedings of the 11th Annual conference on Genetic and evolutionary computation - GECCO '09  
In this paper we prove that for a variety of practical situations, the no-free-lunch (NFL) theorem does not apply to algorithms that search the space of artificial neural networks, such as evolutionary algorithms. We find, in particular, that, while conditions under which NFL applies exist, these require extremely restrictive symmetries on the set of possible problems which are unlikely encountered in practice. In other words, not all algorithms are equally good at finding neural networks that
more » ... ural networks that solve problems under all possible performance measures: a superior search algorithm for this domain does exist.
doi:10.1145/1569901.1570074 dblp:conf/gecco/PoliG09 fatcat:wlwftivakrd5bds5sfxmwx7bre