Matches in SemOpenAlex for { <https://semopenalex.org/work/W3102885980> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W3102885980 abstract "Despite the improvement of translation quality, neural machine translation (NMT) often suffers from the lack of diversity in its generation. In this paper, we propose to generate diverse translations by deriving a large number of possible models with Bayesian modelling and sampling models from them for inference. The possible models are obtained by applying concrete dropout to the NMT model and each of them has specific confidence for its prediction, which corresponds to a posterior model distribution under specific training data in the principle of Bayesian modeling. With variational inference, the posterior model distribution can be approximated with a variational distribution, from which the final models for inference are sampled. We conducted experiments on Chinese-English and English-German translation tasks and the results shows that our method makes a better trade-off between diversity and accuracy." @default.
- W3102885980 created "2020-11-23" @default.
- W3102885980 creator A5000232528 @default.
- W3102885980 creator A5056500539 @default.
- W3102885980 creator A5062962382 @default.
- W3102885980 date "2020-01-01" @default.
- W3102885980 modified "2023-10-16" @default.
- W3102885980 title "Generating Diverse Translation from Model Distribution with Dropout" @default.
- W3102885980 cites W1521968289 @default.
- W3102885980 cites W1567512734 @default.
- W3102885980 cites W1753482797 @default.
- W3102885980 cites W1904365287 @default.
- W3102885980 cites W2047229728 @default.
- W3102885980 cites W2095705004 @default.
- W3102885980 cites W2101105183 @default.
- W3102885980 cites W2108677974 @default.
- W3102885980 cites W2115259925 @default.
- W3102885980 cites W2130942839 @default.
- W3102885980 cites W2183341477 @default.
- W3102885980 cites W2552839021 @default.
- W3102885980 cites W2557436004 @default.
- W3102885980 cites W2752013927 @default.
- W3102885980 cites W2890220768 @default.
- W3102885980 cites W2915573484 @default.
- W3102885980 cites W2933138175 @default.
- W3102885980 cites W2949555952 @default.
- W3102885980 cites W2949938546 @default.
- W3102885980 cites W2951266961 @default.
- W3102885980 cites W2951813108 @default.
- W3102885980 cites W2962784628 @default.
- W3102885980 cites W2963366552 @default.
- W3102885980 cites W2963403868 @default.
- W3102885980 cites W2963684875 @default.
- W3102885980 cites W2964059111 @default.
- W3102885980 cites W2964121744 @default.
- W3102885980 cites W2964345285 @default.
- W3102885980 cites W2970565456 @default.
- W3102885980 cites W2996159613 @default.
- W3102885980 cites W2998547639 @default.
- W3102885980 cites W2998655810 @default.
- W3102885980 doi "https://doi.org/10.18653/v1/2020.emnlp-main.82" @default.
- W3102885980 hasPublicationYear "2020" @default.
- W3102885980 type Work @default.
- W3102885980 sameAs 3102885980 @default.
- W3102885980 citedByCount "4" @default.
- W3102885980 countsByYear W31028859802021 @default.
- W3102885980 countsByYear W31028859802022 @default.
- W3102885980 crossrefType "proceedings-article" @default.
- W3102885980 hasAuthorship W3102885980A5000232528 @default.
- W3102885980 hasAuthorship W3102885980A5056500539 @default.
- W3102885980 hasAuthorship W3102885980A5062962382 @default.
- W3102885980 hasBestOaLocation W31028859801 @default.
- W3102885980 hasConcept C104317684 @default.
- W3102885980 hasConcept C105580179 @default.
- W3102885980 hasConcept C119857082 @default.
- W3102885980 hasConcept C149364088 @default.
- W3102885980 hasConcept C154945302 @default.
- W3102885980 hasConcept C185592680 @default.
- W3102885980 hasConcept C2776145597 @default.
- W3102885980 hasConcept C41008148 @default.
- W3102885980 hasConcept C55493867 @default.
- W3102885980 hasConceptScore W3102885980C104317684 @default.
- W3102885980 hasConceptScore W3102885980C105580179 @default.
- W3102885980 hasConceptScore W3102885980C119857082 @default.
- W3102885980 hasConceptScore W3102885980C149364088 @default.
- W3102885980 hasConceptScore W3102885980C154945302 @default.
- W3102885980 hasConceptScore W3102885980C185592680 @default.
- W3102885980 hasConceptScore W3102885980C2776145597 @default.
- W3102885980 hasConceptScore W3102885980C41008148 @default.
- W3102885980 hasConceptScore W3102885980C55493867 @default.
- W3102885980 hasLocation W31028859801 @default.
- W3102885980 hasLocation W31028859802 @default.
- W3102885980 hasOpenAccess W3102885980 @default.
- W3102885980 hasPrimaryLocation W31028859801 @default.
- W3102885980 hasRelatedWork W1985074889 @default.
- W3102885980 hasRelatedWork W2032126585 @default.
- W3102885980 hasRelatedWork W2068882612 @default.
- W3102885980 hasRelatedWork W2313858059 @default.
- W3102885980 hasRelatedWork W2401401198 @default.
- W3102885980 hasRelatedWork W2555500570 @default.
- W3102885980 hasRelatedWork W3135923011 @default.
- W3102885980 hasRelatedWork W4214508080 @default.
- W3102885980 hasRelatedWork W97426361 @default.
- W3102885980 hasRelatedWork W99879459 @default.
- W3102885980 isParatext "false" @default.
- W3102885980 isRetracted "false" @default.
- W3102885980 magId "3102885980" @default.
- W3102885980 workType "article" @default.