Learning from evolved next release problem instances

Abstract : Taking the Next Release Problem (NRP) as a case study, we intend to analyze the relationship between heuristics and the software engineering problem instances. We adopt an evolutionary algorithm to evolve NRP instances that are either hard or easy for the target heuristic (GRASP in this study), to investigate where a heuristic works well and where it does not, when facing a software engineering problem. Thereafter, we use a feature-based approach to predict the hardness of the evolved instances, with respect to the target heuristic. Experimental results reveal that, the proposed algorithm is able to evolve NRP instances with different hardness. Furthermore, the problem-specific features enables the prediction of the target heuristic's performance.
Type de document :
Communication dans un congrès
GECCO - Genetic and Evolutionary Computation Conference, 2014, Jul 2014, Vancouver, BC, Canada. pp.189 - 190, 2014, 〈10.1145/2598394.2598427〉
Liste complète des métadonnées

Littérature citée [4 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01087436
Contributeur : Jifeng Xuan <>
Soumis le : mercredi 26 novembre 2014 - 10:31:45
Dernière modification le : jeudi 11 janvier 2018 - 06:25:38
Document(s) archivé(s) le : vendredi 27 février 2015 - 11:00:30

Fichier

pap299-he.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

Citation

Zhilei Ren, He Jiang, Jifeng Xuan, Shuwei Zhang, Zhongxuan Luo. Learning from evolved next release problem instances. GECCO - Genetic and Evolutionary Computation Conference, 2014, Jul 2014, Vancouver, BC, Canada. pp.189 - 190, 2014, 〈10.1145/2598394.2598427〉. 〈hal-01087436〉

Partager

Métriques

Consultations de la notice

163

Téléchargements de fichiers

79