Mostra el registre d'ítem simple
Multilingual machine translation: Closing the gap between shared and language-specific encoder-decoders
dc.contributor.author | Escolano Peinado, Carlos |
dc.contributor.author | Ruiz Costa-Jussà, Marta |
dc.contributor.author | Rodríguez Fonollosa, José Adrián |
dc.contributor.author | Artetxe Zurutuza, Mikel |
dc.contributor.other | Universitat Politècnica de Catalunya. Doctorat en Teoria del Senyal i Comunicacions |
dc.contributor.other | Universitat Politècnica de Catalunya. Departament de Ciències de la Computació |
dc.contributor.other | Universitat Politècnica de Catalunya. Departament de Teoria del Senyal i Comunicacions |
dc.date.accessioned | 2021-06-15T08:31:27Z |
dc.date.available | 2021-06-15T08:31:27Z |
dc.date.issued | 2021 |
dc.identifier.citation | Escolano, C. [et al.]. Multilingual machine translation: Closing the gap between shared and language-specific encoder-decoders. A: Conference of the European Chapter of the Association for Computational Linguistics. "EACL 2021, The 16th Conference of the European Chapter of the Association for Computational Linguistics: April 19-23, 2021: proceedings of the conference". Stroudsburg, PA: Association for Computational Linguistics, 2021, p. 944-948. ISBN 978-1-954085-02-2. |
dc.identifier.isbn | 978-1-954085-02-2 |
dc.identifier.uri | http://hdl.handle.net/2117/347326 |
dc.description.abstract | State-of-the-art multilingual machine translation relies on a universal encoder-decoder, which requires retraining the entire system to add new languages. In this paper, we propose an alternative approach that is based on language-specific encoder-decoders, and can thus be more easily extended to new languages by learning their corresponding modules. So as to encourage a common interlingua representation, we simultaneously train the N initial languages. Our experiments show that the proposed approach outperforms the universal encoder-decoder by 3.28 BLEU points on average, while allowing to add new languages without the need to retrain the rest of the modules. All in all, our work closes the gap between shared and language-specific encoderdecoders, advancing toward modular multilingual machine translation systems that can be flexibly extended in lifelong learning settings. |
dc.description.sponsorship | This work is supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No. 947657). |
dc.format.extent | 5 p. |
dc.language.iso | eng |
dc.publisher | Association for Computational Linguistics |
dc.rights | Attribution 4.0 International |
dc.rights.uri | https://creativecommons.org/licenses/by/4.0/ |
dc.subject | Àrees temàtiques de la UPC::Informàtica::Intel·ligència artificial::Llenguatge natural |
dc.subject.lcsh | Machine translating |
dc.subject.other | Multilingual machine translation |
dc.subject.other | Universal encoder-decoder |
dc.subject.other | State-of-the-art |
dc.title | Multilingual machine translation: Closing the gap between shared and language-specific encoder-decoders |
dc.type | Conference lecture |
dc.subject.lemac | Traducció automàtica |
dc.contributor.group | Universitat Politècnica de Catalunya. VEU - Grup de Tractament de la Parla |
dc.description.peerreviewed | Peer Reviewed |
dc.relation.publisherversion | https://www.aclweb.org/anthology/2021.eacl-main.80/ |
dc.rights.access | Open Access |
local.identifier.drac | 30613432 |
dc.description.version | Postprint (published version) |
dc.relation.projectid | info:eu-repo/grantAgreement/EC/H2020/947657/EU/Lifelong UNiversal lAnguage Representation/LUNAR |
local.citation.author | Escolano, C.; Costa-jussà, M. R.; Fonollosa, J. A. R.; Artetxe, M. |
local.citation.contributor | Conference of the European Chapter of the Association for Computational Linguistics |
local.citation.pubplace | Stroudsburg, PA |
local.citation.publicationName | EACL 2021, The 16th Conference of the European Chapter of the Association for Computational Linguistics: April 19-23, 2021: proceedings of the conference |
local.citation.startingPage | 944 |
local.citation.endingPage | 948 |