Evolutionary Algorithms as fitness function debuggers - Inria - Institut national de recherche en sciences et technologies du numérique
Chapitre D'ouvrage Année : 1999

Evolutionary Algorithms as fitness function debuggers

Résumé

All Evolutionary Algorithms experienced practitioners emphasize the need for a careful design of the fitness function. It is commonly heard~ for instance, that "If there is a bug in your fitness function, the EA will find it". This paper presents a case study of such a situation in the domain of geophysical underground identification: some weird solutions are found by the Evolutionary Algorithm, obviously physically absurd, but fulfilling almost perfectly the geophysical criterion.
Fichier principal
Vignette du fichier
debugger-FIS99.pdf (801.31 Ko) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte
Loading...

Dates et versions

hal-01225296 , version 1 (09-11-2015)

Identifiants

Citer

Frédéric Mansanne, Frédéric Carrère, Andréas Ehinger, Marc Schoenauer. Evolutionary Algorithms as fitness function debuggers. Raś, ZbigniewW. and Skowron, Andrzej. Foundations of Intelligent Systems, 1609, Springer Verlag, pp.639-647, 1999, Lecture Notes in Computer Science, 978-3-540-65965-5. ⟨10.1007/BFb00910.1007/BFb00951535153⟩. ⟨hal-01225296⟩
214 Consultations
629 Téléchargements

Altmetric

Partager

More