ActivMetaL: Algorithm Recommendation with Active Meta Learning

Lisheng Sun-Hosoya 1 Isabelle Guyon 1 Michèle Sebag 2
1 TAU - TAckling the Underspecified
LRI - Laboratoire de Recherche en Informatique, Inria Saclay - Ile de France, CNRS - Centre National de la Recherche Scientifique : UMR8623, UP11 - Université Paris-Sud - Paris 11
Abstract : We present an active meta learning approach to model selection or algorithm recommendation. We adopt the point of view "collab-orative filtering" recommender systems in which the problem is brought back to a missing data problem: given a sparsely populated matrix of performances of algorithms on given tasks, predict missing performances; more particularly, predict which algorithm will perform best on a new dataset (empty row). In this work, we propose and study an active learning version of the recommender algorithm CofiRank algorithm and compare it with baseline methods. Our benchmark involves three real-world datasets (from StatLog, OpenML, and AutoML) and artificial data. Our results indicate that CofiRank rapidly finds well performing algorithms on new datasets at reasonable computational cost.
Type de document :
Poster
IAL 2018 workshop, ECML PKDD, Sep 2018, Dublin, Ireland. 2018, 〈http://ceur-ws.org/Vol-2192/ialatecml_paper3.pdf〉
Liste complète des métadonnées

Littérature citée [12 références]  Voir  Masquer  Télécharger

https://hal.archives-ouvertes.fr/hal-01931262
Contributeur : Lisheng Sun-Hosoya <>
Soumis le : jeudi 22 novembre 2018 - 16:59:52
Dernière modification le : jeudi 7 février 2019 - 16:55:42
Document(s) archivé(s) le : samedi 23 février 2019 - 15:23:09

Fichier

ACTIVMETAL.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01931262, version 1

Citation

Lisheng Sun-Hosoya, Isabelle Guyon, Michèle Sebag. ActivMetaL: Algorithm Recommendation with Active Meta Learning. IAL 2018 workshop, ECML PKDD, Sep 2018, Dublin, Ireland. 2018, 〈http://ceur-ws.org/Vol-2192/ialatecml_paper3.pdf〉. 〈hal-01931262〉

Partager

Métriques

Consultations de la notice

65

Téléchargements de fichiers

111