Matches in SemOpenAlex for { <https://semopenalex.org/work/W2997284037> ?p ?o ?g. }
Showing items 1 to 93 of
93
with 100 items per page.
- W2997284037 endingPage "205" @default.
- W2997284037 startingPage "198" @default.
- W2997284037 abstract "Non-Autoregressive Neural Machine Translation (NAT) achieves significant decoding speedup through generating target words independently and simultaneously. However, in the context of non-autoregressive translation, the word-level cross-entropy loss cannot model the target-side sequential dependency properly, leading to its weak correlation with the translation quality. As a result, NAT tends to generate influent translations with over-translation and under-translation errors. In this paper, we propose to train NAT to minimize the Bag-of-Ngrams (BoN) difference between the model output and the reference sentence. The bag-of-ngrams training objective is differentiable and can be efficiently calculated, which encourages NAT to capture the target-side sequential dependency and correlates well with the translation quality. We validate our approach on three translation tasks and show that our approach largely outperforms the NAT baseline by about 5.0 BLEU scores on WMT14 En↔De and about 2.5 BLEU scores on WMT16 En↔Ro." @default.
- W2997284037 created "2020-01-10" @default.
- W2997284037 creator A5000232528 @default.
- W2997284037 creator A5024849044 @default.
- W2997284037 creator A5056500539 @default.
- W2997284037 creator A5066557235 @default.
- W2997284037 creator A5076799733 @default.
- W2997284037 date "2020-04-03" @default.
- W2997284037 modified "2023-10-14" @default.
- W2997284037 title "Minimizing the Bag-of-Ngrams Difference for Non-Autoregressive Neural Machine Translation" @default.
- W2997284037 doi "https://doi.org/10.1609/aaai.v34i01.5351" @default.
- W2997284037 hasPublicationYear "2020" @default.
- W2997284037 type Work @default.
- W2997284037 sameAs 2997284037 @default.
- W2997284037 citedByCount "13" @default.
- W2997284037 countsByYear W29972840372021 @default.
- W2997284037 countsByYear W29972840372022 @default.
- W2997284037 countsByYear W29972840372023 @default.
- W2997284037 crossrefType "journal-article" @default.
- W2997284037 hasAuthorship W2997284037A5000232528 @default.
- W2997284037 hasAuthorship W2997284037A5024849044 @default.
- W2997284037 hasAuthorship W2997284037A5056500539 @default.
- W2997284037 hasAuthorship W2997284037A5066557235 @default.
- W2997284037 hasAuthorship W2997284037A5076799733 @default.
- W2997284037 hasBestOaLocation W29972840371 @default.
- W2997284037 hasConcept C104317684 @default.
- W2997284037 hasConcept C105580179 @default.
- W2997284037 hasConcept C105795698 @default.
- W2997284037 hasConcept C11413529 @default.
- W2997284037 hasConcept C119857082 @default.
- W2997284037 hasConcept C149364088 @default.
- W2997284037 hasConcept C151730666 @default.
- W2997284037 hasConcept C154945302 @default.
- W2997284037 hasConcept C159877910 @default.
- W2997284037 hasConcept C182516595 @default.
- W2997284037 hasConcept C185592680 @default.
- W2997284037 hasConcept C19768560 @default.
- W2997284037 hasConcept C203005215 @default.
- W2997284037 hasConcept C204321447 @default.
- W2997284037 hasConcept C2777530160 @default.
- W2997284037 hasConcept C2779343474 @default.
- W2997284037 hasConcept C28490314 @default.
- W2997284037 hasConcept C31258907 @default.
- W2997284037 hasConcept C33923547 @default.
- W2997284037 hasConcept C41008148 @default.
- W2997284037 hasConcept C55493867 @default.
- W2997284037 hasConcept C57273362 @default.
- W2997284037 hasConcept C86803240 @default.
- W2997284037 hasConceptScore W2997284037C104317684 @default.
- W2997284037 hasConceptScore W2997284037C105580179 @default.
- W2997284037 hasConceptScore W2997284037C105795698 @default.
- W2997284037 hasConceptScore W2997284037C11413529 @default.
- W2997284037 hasConceptScore W2997284037C119857082 @default.
- W2997284037 hasConceptScore W2997284037C149364088 @default.
- W2997284037 hasConceptScore W2997284037C151730666 @default.
- W2997284037 hasConceptScore W2997284037C154945302 @default.
- W2997284037 hasConceptScore W2997284037C159877910 @default.
- W2997284037 hasConceptScore W2997284037C182516595 @default.
- W2997284037 hasConceptScore W2997284037C185592680 @default.
- W2997284037 hasConceptScore W2997284037C19768560 @default.
- W2997284037 hasConceptScore W2997284037C203005215 @default.
- W2997284037 hasConceptScore W2997284037C204321447 @default.
- W2997284037 hasConceptScore W2997284037C2777530160 @default.
- W2997284037 hasConceptScore W2997284037C2779343474 @default.
- W2997284037 hasConceptScore W2997284037C28490314 @default.
- W2997284037 hasConceptScore W2997284037C31258907 @default.
- W2997284037 hasConceptScore W2997284037C33923547 @default.
- W2997284037 hasConceptScore W2997284037C41008148 @default.
- W2997284037 hasConceptScore W2997284037C55493867 @default.
- W2997284037 hasConceptScore W2997284037C57273362 @default.
- W2997284037 hasConceptScore W2997284037C86803240 @default.
- W2997284037 hasIssue "01" @default.
- W2997284037 hasLocation W29972840371 @default.
- W2997284037 hasLocation W29972840372 @default.
- W2997284037 hasOpenAccess W2997284037 @default.
- W2997284037 hasPrimaryLocation W29972840371 @default.
- W2997284037 hasRelatedWork W170118632 @default.
- W2997284037 hasRelatedWork W1978622157 @default.
- W2997284037 hasRelatedWork W1978971213 @default.
- W2997284037 hasRelatedWork W2507994462 @default.
- W2997284037 hasRelatedWork W2606032440 @default.
- W2997284037 hasRelatedWork W3107474891 @default.
- W2997284037 hasRelatedWork W33032091 @default.
- W2997284037 hasRelatedWork W34164803 @default.
- W2997284037 hasRelatedWork W4298354962 @default.
- W2997284037 hasRelatedWork W4310560507 @default.
- W2997284037 hasVolume "34" @default.
- W2997284037 isParatext "false" @default.
- W2997284037 isRetracted "false" @default.
- W2997284037 magId "2997284037" @default.
- W2997284037 workType "article" @default.