Skip to Main content Skip to Navigation
New interface
Journal articles

Universal Consistency and Bloat in GP

Sylvain Gelly 1 Olivier Teytaud 1 Nicolas Bredeche 1 Marc Schoenauer 1 
1 TANC - Algorithmic number theory for cryptology
LIX - Laboratoire d'informatique de l'École polytechnique [Palaiseau], Inria Saclay - Ile de France
Abstract : In this paper, we provide an analysis of Genetic Programming (GP) from the Statistical Learning Theory viewpoint in the scope of symbolic regression. Firstly, we are interested in Universal Consistency, i.e. the fact that the solution minimizing the empirical error does converge to the best possible error when the number of examples goes to infinity, and secondly, we focus our attention on the uncontrolled growth of program length (i.e. bloat), which is a well-known problem in GP. Results show that (1) several kinds of code bloats may be identified and that (2) Universal consistency can be obtained as well as avoiding bloat under some con- ditions. We conclude by describing an ad hoc method that makes it possible simultaneously to avoid bloat and to ensure universal consistency.
Complete list of metadata

Cited literature [18 references]  Display  Hide  Download
Contributor : Sylvain Gelly Connect in order to contact the contributor
Submitted on : Friday, November 10, 2006 - 10:31:29 AM
Last modification on : Friday, May 13, 2022 - 11:02:03 AM
Long-term archiving on: : Thursday, September 20, 2012 - 2:35:49 PM


  • HAL Id : inria-00112840, version 1



Sylvain Gelly, Olivier Teytaud, Nicolas Bredeche, Marc Schoenauer. Universal Consistency and Bloat in GP. Revue des Sciences et Technologies de l'Information - Série RIA : Revue d'Intelligence Artificielle, 2006. ⟨inria-00112840⟩



Record views


Files downloads