GAWO: Genetic-based optimization algorithm for SMT

Ameur Douib 1 David Langlois 2, 3 Kamel Smaili 1
1 SMarT - Statistical Machine Translation and Speech Modelization and Text
LORIA - NLPKD - Department of Natural Language Processing & Knowledge Discovery
3 APC - THEORIE
APC - UMR 7164 - AstroParticule et Cosmologie, Institut für theoretische Physik
Abstract : In this work, we propose GAWO, a new method for SMT parameters optimization based on the genetic algorithms. Like other existing methods, GAWO performs the optimization task through two nested loops, one for the translation and the other for the optimization. However, our proposition is especially designed to optimize the feature weights of the fitness function of GAMaT, a new genetic-based decoder for SMT. We tested GAWO to optimize GAMaT for French-English and Turkish-English translation tasks, and the results showed that we out-perform the previous performance by +4.0 points according to the BLEU for French-English and by +2.2 points for Turkish-English.
Type de document :
Communication dans un congrès
ICNLSSP 2017 - International Conference on Natural Language, Signal and Speech Processing, Dec 2017, Maroc, Morocco. 2017, 〈http://icnlssp.isga.ma〉
Liste complète des métadonnées

Littérature citée [17 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01660010
Contributeur : Kamel Smaïli <>
Soumis le : samedi 9 décembre 2017 - 17:49:48
Dernière modification le : jeudi 1 février 2018 - 01:36:21

Fichier

ICNLSSP2017_paper_18.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01660010, version 1

Citation

Ameur Douib, David Langlois, Kamel Smaili. GAWO: Genetic-based optimization algorithm for SMT. ICNLSSP 2017 - International Conference on Natural Language, Signal and Speech Processing, Dec 2017, Maroc, Morocco. 2017, 〈http://icnlssp.isga.ma〉. 〈hal-01660010〉

Partager

Métriques

Consultations de la notice

420

Téléchargements de fichiers

32