FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2952614664> ?p ?o ?g. }

• W2952614664 abstract "Transfer learning or multilingual model is essential for low-resource neural machine translation (NMT), but the applicability is limited to cognate languages by sharing their vocabularies. This paper shows effective techniques to transfer a pretrained NMT model to a new, unrelated language without shared vocabularies. We relieve the vocabulary mismatch by using cross-lingual word embedding, train a more language-agnostic encoder by injecting artificial noises, and generate synthetic data easily from the pretraining data without back-translation. Our methods do not require restructuring the vocabulary or retraining the model. We improve plain NMT transfer by up to +5.1% BLEU in five low-resource translation tasks, outperforming multilingual joint training by a large margin. We also provide extensive ablation studies on pretrained embedding, synthetic data, vocabulary size, and parameter freezing for a better understanding of NMT transfer." @default.
• W2952614664 created "2019-06-27" @default.
• W2952614664 creator A5005673540 @default.
• W2952614664 creator A5046577673 @default.
• W2952614664 creator A5087367411 @default.
• W2952614664 date "2019-01-01" @default.
• W2952614664 modified "2023-09-23" @default.
• W2952614664 title "Effective Cross-lingual Transfer of Neural Machine Translation Models without Shared Vocabularies" @default.
• W2952614664 cites W171476473 @default.
• W2952614664 cites W2119727789 @default.
• W2952614664 cites W2123442489 @default.
• W2952614664 cites W2124202134 @default.
• W2952614664 cites W2130942839 @default.
• W2952614664 cites W2155732278 @default.
• W2952614664 cites W2168939893 @default.
• W2952614664 cites W2251743902 @default.
• W2952614664 cites W2295582178 @default.
• W2952614664 cites W2493916176 @default.
• W2952614664 cites W2550821151 @default.
• W2952614664 cites W2555428947 @default.
• W2952614664 cites W2567571499 @default.
• W2952614664 cites W2610245951 @default.
• W2952614664 cites W2741602058 @default.
• W2952614664 cites W2756566411 @default.
• W2952614664 cites W2778814079 @default.
• W2952614664 cites W2887920589 @default.
• W2952614664 cites W2888456631 @default.
• W2952614664 cites W2889894161 @default.
• W2952614664 cites W2890007195 @default.
• W2952614664 cites W2890731353 @default.
• W2952614664 cites W2891896107 @default.
• W2952614664 cites W2902698002 @default.
• W2952614664 cites W2913659301 @default.
• W2952614664 cites W2914120296 @default.
• W2952614664 cites W2950577311 @default.
• W2952614664 cites W2955541912 @default.
• W2952614664 cites W2962784628 @default.
• W2952614664 cites W2962801832 @default.
• W2952614664 cites W2962824887 @default.
• W2952614664 cites W2963088995 @default.
• W2952614664 cites W2963118869 @default.
• W2952614664 cites W2963216553 @default.
• W2952614664 cites W2963247703 @default.
• W2952614664 cites W2963403868 @default.
• W2952614664 cites W2963506925 @default.
• W2952614664 cites W2963602293 @default.
• W2952614664 cites W2963633299 @default.
• W2952614664 cites W2963804993 @default.
• W2952614664 cites W2963831310 @default.
• W2952614664 cites W2963983698 @default.
• W2952614664 cites W2963993537 @default.
• W2952614664 cites W2964085268 @default.
• W2952614664 cites W2964121744 @default.
• W2952614664 cites W2964265128 @default.
• W2952614664 cites W2964266061 @default.
• W2952614664 cites W2964308564 @default.
• W2952614664 cites W3041866211 @default.
• W2952614664 cites W3104652516 @default.
• W2952614664 cites W3204406378 @default.
• W2952614664 cites W32870568 @default.
• W2952614664 cites W99485931 @default.
• W2952614664 doi "https://doi.org/10.18653/v1/p19-1120" @default.
• W2952614664 hasPublicationYear "2019" @default.
• W2952614664 type Work @default.
• W2952614664 sameAs 2952614664 @default.
• W2952614664 citedByCount "48" @default.
• W2952614664 countsByYear W29526146642019 @default.
• W2952614664 countsByYear W29526146642020 @default.
• W2952614664 countsByYear W29526146642021 @default.
• W2952614664 countsByYear W29526146642022 @default.
• W2952614664 countsByYear W29526146642023 @default.
• W2952614664 crossrefType "proceedings-article" @default.
• W2952614664 hasAuthorship W2952614664A5005673540 @default.
• W2952614664 hasAuthorship W2952614664A5046577673 @default.
• W2952614664 hasAuthorship W2952614664A5087367411 @default.
• W2952614664 hasBestOaLocation W29526146641 @default.
• W2952614664 hasConcept C104317684 @default.
• W2952614664 hasConcept C105580179 @default.
• W2952614664 hasConcept C111919701 @default.
• W2952614664 hasConcept C118505674 @default.
• W2952614664 hasConcept C119857082 @default.
• W2952614664 hasConcept C137293760 @default.
• W2952614664 hasConcept C138885662 @default.
• W2952614664 hasConcept C144133560 @default.
• W2952614664 hasConcept C149364088 @default.
• W2952614664 hasConcept C150899416 @default.
• W2952614664 hasConcept C154945302 @default.
• W2952614664 hasConcept C155202549 @default.
• W2952614664 hasConcept C173608175 @default.
• W2952614664 hasConcept C185592680 @default.
• W2952614664 hasConcept C203005215 @default.
• W2952614664 hasConcept C204321447 @default.
• W2952614664 hasConcept C2776175482 @default.
• W2952614664 hasConcept C2777601683 @default.
• W2952614664 hasConcept C2778712577 @default.
• W2952614664 hasConcept C28490314 @default.
• W2952614664 hasConcept C41008148 @default.
• W2952614664 hasConcept C41608201 @default.
• W2952614664 hasConcept C41895202 @default.
• W2952614664 hasConcept C55493867 @default.

• next