Matches in SemOpenAlex for { <https://semopenalex.org/work/W3175095087> ?p ?o ?g. }
- W3175095087 endingPage "12479" @default.
- W3175095087 startingPage "12471" @default.
- W3175095087 abstract "UNMT tackles translation on monolingual corpora in two required languages. Since there is no explicitly cross-lingual signal, pre-training and synthetic sentence pairs are significant to the success of UNMT. In this work, we empirically study the core training procedure of UNMT to analyze the synthetic sentence pairs obtained from back-translation. We introduce new losses to UNMT to regularize the synthetic sentence pairs by jointly training the UNMT objective and the regularization objective. Our comprehensive experiments support that our method can generally improve the performance of currently successful models on three similar pairs {French, German, Romanian} English and one dissimilar pair Russian English with acceptably additional cost." @default.
- W3175095087 created "2021-07-05" @default.
- W3175095087 creator A5046520540 @default.
- W3175095087 creator A5081756234 @default.
- W3175095087 date "2021-05-18" @default.
- W3175095087 modified "2023-10-16" @default.
- W3175095087 title "Empirical Regularization for Synthetic Sentence Pairs in Unsupervised Neural Machine Translation" @default.
- W3175095087 cites W2099471712 @default.
- W3175095087 cites W2130942839 @default.
- W3175095087 cites W2145094598 @default.
- W3175095087 cites W2153579005 @default.
- W3175095087 cites W2153653739 @default.
- W3175095087 cites W2170973209 @default.
- W3175095087 cites W2402144811 @default.
- W3175095087 cites W2493916176 @default.
- W3175095087 cites W2525778437 @default.
- W3175095087 cites W2561995736 @default.
- W3175095087 cites W2740132093 @default.
- W3175095087 cites W2740534027 @default.
- W3175095087 cites W2741602058 @default.
- W3175095087 cites W2885023989 @default.
- W3175095087 cites W2914120296 @default.
- W3175095087 cites W2944815030 @default.
- W3175095087 cites W2949405462 @default.
- W3175095087 cites W2950428495 @default.
- W3175095087 cites W2952614664 @default.
- W3175095087 cites W2962784628 @default.
- W3175095087 cites W2962824887 @default.
- W3175095087 cites W2963118869 @default.
- W3175095087 cites W2963206679 @default.
- W3175095087 cites W2963216553 @default.
- W3175095087 cites W2963223306 @default.
- W3175095087 cites W2963341956 @default.
- W3175095087 cites W2963413917 @default.
- W3175095087 cites W2963443683 @default.
- W3175095087 cites W2963602293 @default.
- W3175095087 cites W2963804993 @default.
- W3175095087 cites W2964013027 @default.
- W3175095087 cites W2964076537 @default.
- W3175095087 cites W2964121744 @default.
- W3175095087 cites W2964308564 @default.
- W3175095087 cites W2982399380 @default.
- W3175095087 cites W2995015695 @default.
- W3175095087 cites W3030163527 @default.
- W3175095087 cites W3107826490 @default.
- W3175095087 cites W8895266 @default.
- W3175095087 doi "https://doi.org/10.1609/aaai.v35i14.17479" @default.
- W3175095087 hasPublicationYear "2021" @default.
- W3175095087 type Work @default.
- W3175095087 sameAs 3175095087 @default.
- W3175095087 citedByCount "0" @default.
- W3175095087 crossrefType "journal-article" @default.
- W3175095087 hasAuthorship W3175095087A5046520540 @default.
- W3175095087 hasAuthorship W3175095087A5081756234 @default.
- W3175095087 hasBestOaLocation W31750950871 @default.
- W3175095087 hasConcept C104317684 @default.
- W3175095087 hasConcept C105580179 @default.
- W3175095087 hasConcept C129400051 @default.
- W3175095087 hasConcept C138885662 @default.
- W3175095087 hasConcept C149364088 @default.
- W3175095087 hasConcept C154775046 @default.
- W3175095087 hasConcept C154945302 @default.
- W3175095087 hasConcept C160920958 @default.
- W3175095087 hasConcept C185592680 @default.
- W3175095087 hasConcept C203005215 @default.
- W3175095087 hasConcept C204321447 @default.
- W3175095087 hasConcept C2776135515 @default.
- W3175095087 hasConcept C2777530160 @default.
- W3175095087 hasConcept C28490314 @default.
- W3175095087 hasConcept C41008148 @default.
- W3175095087 hasConcept C41895202 @default.
- W3175095087 hasConcept C51632099 @default.
- W3175095087 hasConcept C55493867 @default.
- W3175095087 hasConceptScore W3175095087C104317684 @default.
- W3175095087 hasConceptScore W3175095087C105580179 @default.
- W3175095087 hasConceptScore W3175095087C129400051 @default.
- W3175095087 hasConceptScore W3175095087C138885662 @default.
- W3175095087 hasConceptScore W3175095087C149364088 @default.
- W3175095087 hasConceptScore W3175095087C154775046 @default.
- W3175095087 hasConceptScore W3175095087C154945302 @default.
- W3175095087 hasConceptScore W3175095087C160920958 @default.
- W3175095087 hasConceptScore W3175095087C185592680 @default.
- W3175095087 hasConceptScore W3175095087C203005215 @default.
- W3175095087 hasConceptScore W3175095087C204321447 @default.
- W3175095087 hasConceptScore W3175095087C2776135515 @default.
- W3175095087 hasConceptScore W3175095087C2777530160 @default.
- W3175095087 hasConceptScore W3175095087C28490314 @default.
- W3175095087 hasConceptScore W3175095087C41008148 @default.
- W3175095087 hasConceptScore W3175095087C41895202 @default.
- W3175095087 hasConceptScore W3175095087C51632099 @default.
- W3175095087 hasConceptScore W3175095087C55493867 @default.
- W3175095087 hasIssue "14" @default.
- W3175095087 hasLocation W31750950871 @default.
- W3175095087 hasOpenAccess W3175095087 @default.
- W3175095087 hasPrimaryLocation W31750950871 @default.
- W3175095087 hasRelatedWork W170118632 @default.
- W3175095087 hasRelatedWork W1947252915 @default.
- W3175095087 hasRelatedWork W1978971213 @default.