Multilingual Neural Machine Translation for Low-Resource Languages

In recent years, Neural Machine Translation (NMT) has been shown to be more effective than phrase-based statistical methods, thus quickly becoming the state of the art in machine translation (MT). However, NMT systems are limited in translating low-resourced languages, due to the significant amount...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Surafel M. Lakew, Marcello Federico, Matteo Negri, Marco Turchi
Formato: article
Lenguaje:EN
Publicado: Accademia University Press 2018
Materias:
H
Acceso en línea:https://doaj.org/article/2b007dd0adc846848c6d4c6ca15810d3
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:2b007dd0adc846848c6d4c6ca15810d3
record_format dspace
spelling oai:doaj.org-article:2b007dd0adc846848c6d4c6ca15810d32021-12-02T09:52:26ZMultilingual Neural Machine Translation for Low-Resource Languages2499-455310.4000/ijcol.531https://doaj.org/article/2b007dd0adc846848c6d4c6ca15810d32018-06-01T00:00:00Zhttp://journals.openedition.org/ijcol/531https://doaj.org/toc/2499-4553In recent years, Neural Machine Translation (NMT) has been shown to be more effective than phrase-based statistical methods, thus quickly becoming the state of the art in machine translation (MT). However, NMT systems are limited in translating low-resourced languages, due to the significant amount of parallel data that is required to learn useful mappings between languages. In this work, we show how the so-called multilingual NMT can help to tackle the challenges associated with low-resourced language translation. The underlying principle of multilingual NMT is to force the creation of hidden representations of words in a shared semantic space across multiple languages, thus enabling a positive parameter transfer across languages. Along this direction, we present multilingual translation experiments with three languages (English, Italian, Romanian) covering six translation directions, utilizing both recurrent neural networks and transformer (or self-attentive) neural networks. We then focus on the zero-shot translation problem, that is how to leverage multi-lingual data in order to learn translation directions that are not covered by the available training material. To this aim, we introduce our recently proposed iterative self-training method, which incrementally improves a multilingual NMT on a zero-shot direction by just relying on monolingual data. Our results on TED talks data show that multilingual NMT outperforms conventional bilingual NMT, that the transformer NMT outperforms recurrent NMT, and that zero-shot NMT outperforms conventional pivoting methods and even matches the performance of a fully-trained bilingual system.Surafel M. LakewMarcello FedericoMatteo NegriMarco TurchiAccademia University PressarticleSocial SciencesHComputational linguistics. Natural language processingP98-98.5ENIJCoL, Vol 4, Iss 1, Pp 11-25 (2018)
institution DOAJ
collection DOAJ
language EN
topic Social Sciences
H
Computational linguistics. Natural language processing
P98-98.5
spellingShingle Social Sciences
H
Computational linguistics. Natural language processing
P98-98.5
Surafel M. Lakew
Marcello Federico
Matteo Negri
Marco Turchi
Multilingual Neural Machine Translation for Low-Resource Languages
description In recent years, Neural Machine Translation (NMT) has been shown to be more effective than phrase-based statistical methods, thus quickly becoming the state of the art in machine translation (MT). However, NMT systems are limited in translating low-resourced languages, due to the significant amount of parallel data that is required to learn useful mappings between languages. In this work, we show how the so-called multilingual NMT can help to tackle the challenges associated with low-resourced language translation. The underlying principle of multilingual NMT is to force the creation of hidden representations of words in a shared semantic space across multiple languages, thus enabling a positive parameter transfer across languages. Along this direction, we present multilingual translation experiments with three languages (English, Italian, Romanian) covering six translation directions, utilizing both recurrent neural networks and transformer (or self-attentive) neural networks. We then focus on the zero-shot translation problem, that is how to leverage multi-lingual data in order to learn translation directions that are not covered by the available training material. To this aim, we introduce our recently proposed iterative self-training method, which incrementally improves a multilingual NMT on a zero-shot direction by just relying on monolingual data. Our results on TED talks data show that multilingual NMT outperforms conventional bilingual NMT, that the transformer NMT outperforms recurrent NMT, and that zero-shot NMT outperforms conventional pivoting methods and even matches the performance of a fully-trained bilingual system.
format article
author Surafel M. Lakew
Marcello Federico
Matteo Negri
Marco Turchi
author_facet Surafel M. Lakew
Marcello Federico
Matteo Negri
Marco Turchi
author_sort Surafel M. Lakew
title Multilingual Neural Machine Translation for Low-Resource Languages
title_short Multilingual Neural Machine Translation for Low-Resource Languages
title_full Multilingual Neural Machine Translation for Low-Resource Languages
title_fullStr Multilingual Neural Machine Translation for Low-Resource Languages
title_full_unstemmed Multilingual Neural Machine Translation for Low-Resource Languages
title_sort multilingual neural machine translation for low-resource languages
publisher Accademia University Press
publishDate 2018
url https://doaj.org/article/2b007dd0adc846848c6d4c6ca15810d3
work_keys_str_mv AT surafelmlakew multilingualneuralmachinetranslationforlowresourcelanguages
AT marcellofederico multilingualneuralmachinetranslationforlowresourcelanguages
AT matteonegri multilingualneuralmachinetranslationforlowresourcelanguages
AT marcoturchi multilingualneuralmachinetranslationforlowresourcelanguages
_version_ 1718397934454702080