LIUM Machine Translation Systems for WMT17 News Translation Task

This paper describes LIUM submissions to WMT17 News Translation Task for English↔German, English↔Turkish, English→Czech and English→Latvian language pairs. We train BPE-based attentive Neural Machine Translation systems with and without factored outputs using the open source nmtpy framework. Competitive scores were obtained by en-sembling various systems and exploiting the availability of target monolingual corpora for back-translation. The impact of back-translation quantity and quality is also analyzed for English→Turkish where our post-deadline submission surpassed the best entry by +1.6 BLEU.

Domaines

Réseau de neurones [cs.NE] Informatique et langage [cs.CL] Machine Learning [stat.ML]

Fichier principal

paper.pdf (180.2 Ko)

Origine	Fichiers produits par l'(les) auteur(s)

Mercedes Garcia-Martinez : Connectez-vous pour contacter le contributeur

https://univ-lemans.hal.science/hal-01742378

Soumis le : lundi 26 mars 2018-11:36:51

Dernière modification le : mercredi 3 juillet 2024-11:54:03

Archivage à long terme le : jeudi 13 septembre 2018-08:05:04

Dates et versions

hal-01742378 , version 1 (26-03-2018)

Identifiants

HAL Id : hal-01742378 , version 1

Citer

Mercedes Garcia-Martinez, Ozan Caglayan, Walid Aransa, Adrien Bardet, Fethi Bougares, et al.. LIUM Machine Translation Systems for WMT17 News Translation Task. Second Conference on Machine Translation, 2017, Copenhagen, Denmark. pp.288 - 295. ⟨hal-01742378⟩

Exporter

BibTeX XML-TEI Dublin Core DC Terms EndNote DataCite

Collections

UNIV-LEMANS LIUM LIUM-LST

157 Consultations

167 Téléchargements