An Evaluation of Language-Agnostic Inner-Attention-Based Representations in Machine Translation

Visa fullständig post



Permalänk

http://hdl.handle.net/10138/305136

Citation

Raganato , A , Vázquez , R , Creutz , M & Tiedemann , J 2019 , An Evaluation of Language-Agnostic Inner-Attention-Based Representations in Machine Translation . in I Augenstein , S Gella , S Ruder , K Kann , B Can , J Welbl , A Conneau , X Ren & M Rei (eds) , The 4th Workshop on Representation Learning for NLP (RepL4NLP-2019) : Proceedings of the Workshop . The Association for Computational Linguistics , Stroudsburg , pp. 27-32 , Workshop on Representation Learning for NLP , Florence , Italy , 02/08/2019 . < https://www.aclweb.org/anthology/W19-4304 >

Titel: An Evaluation of Language-Agnostic Inner-Attention-Based Representations in Machine Translation
Författare: Raganato, Alessandro; Vázquez, Raúl; Creutz, Mathias; Tiedemann, Jörg
Medarbetare: Augenstein, Isabelle
Gella, Spandana
Ruder, Sebastian
Kann, Katharina
Can, Burcu
Welbl, Johannes
Conneau, Alexis
Ren, Xiang
Rei, Marek
Upphovmannens organisation: Department of Digital Humanities
Language Technology
Utgivare: The Association for Computational Linguistics
Datum: 2019-08-01
Språk: eng
Sidantal: 6
Tillhör serie: The 4th Workshop on Representation Learning for NLP (RepL4NLP-2019)
ISBN: 978-1-950737-35-2
Permanenta länken (URI): http://hdl.handle.net/10138/305136
Abstrakt: In this paper, we explore a multilingual translation model with a cross-lingually shared layer that can be used as fixed-size sentence representation in different downstream tasks. We systematically study the impact of the size of the shared layer and the effect of including additional languages in the model. In contrast to related previous work, we demonstrate that the performance in translation does correlate with trainable downstream tasks. In particular, we show that larger intermediate layers not only improve translation quality, especially for long sentences, but also push the accuracy of trainable classification tasks. On the other hand, shorter representations lead to increased compression that is beneficial in non-trainable similarity tasks. We hypothesize that the training procedure on the downstream task enables the model to identify the encoded information that is useful for the specific task whereas non-trainable benchmarks can be confused by other types of information also encoded in the representation of a sentence.
Subject: 6121 Languages
113 Computer and information sciences
Referentgranskad: Ja
Licens: cc_by
Användningsbegränsning: openAccess
Parallelpublicerad version: publishedVersion
Finansierad av: European Commission
SUOMEN AKATEMIA
Finansierings ID: 771113


Filer under denna titel

Totalt antal nerladdningar: Laddar...

Filer Storlek Format Granska
W19_4304.pdf 371.9Kb PDF Granska/Öppna

Detta dokument registreras i samling:

Visa fullständig post