Graduate School and Research Center in Digital Sciences

The MeMAD submission to the WMT18 multimodal translation task

Grönroos, Stig-Arne; Huet, Benoit; Kurimo, Mikko; Laaksonen, Jorma; Merialdo, Bernard; Pham, Phu; Sjöberg, Mats; Sulubacak, Umut; Tiedemann, Jörg; Troncy, Raphael; Vázquez, Raúl

WMT 2018, 3rd Conference on Machine Translation, October 31-November 1st, 2018, Brussels, Belgium

This paper describes the MeMAD project entry to the WMT Multimodal Machine Translation Shared Task. We propose adapting the Transformer neural machine translation (NMT) architecture to a multi-modal setting. In this paper, we also describe the preliminary experiments with textonly translation systems leading us up to this choice. We have the top scoring system for both English-to-German and English-to-French, according to the automatic metrics for flickr18. Our experiments show that the effect of the visual features in our system is small. Our largest gains come from the quality of the underlying text-only NMT system. We find that appropriate use of additional data is effective.

Document Arxiv Bibtex

Title:The MeMAD submission to the WMT18 multimodal translation task
Department:Data Science
Eurecom ref:5665
Bibtex: @inproceedings{EURECOM+5665, year = {2018}, title = {{T}he {M}e{MAD} submission to the {WMT}18 multimodal translation task}, author = {{G}r{\"o}nroos, {S}tig-{A}rne and {H}uet, {B}enoit and {K}urimo, {M}ikko and {L}aaksonen, {J}orma and {M}erialdo, {B}ernard and {P}ham, {P}hu and {S}j{\"o}berg, {M}ats and {S}ulubacak, {U}mut and {T}iedemann, {J}{\"o}rg and {T}roncy, {R}aphael and {V}{\'a}zquez, {R}a{\'u}l}, booktitle = {{WMT} 2018, 3rd {C}onference on {M}achine {T}ranslation, {O}ctober 31-{N}ovember 1st, 2018, {B}russels, {B}elgium}, address = {{B}russels, {BELGIUM}}, month = {10}, url = {} }
See also: