Matches in SemOpenAlex for { <https://semopenalex.org/work/W3008125272> ?p ?o ?g. }
- W3008125272 abstract "Nowadays, training end-to-end neural models for spoken language translation (SLT) still has to confront with extreme data scarcity conditions. The existing SLT parallel corpora are indeed orders of magnitude smaller than those available for the closely related tasks of automatic speech recognition (ASR) and machine translation (MT), which usually comprise tens of millions of instances. To cope with data paucity, in this paper we explore the effectiveness of transfer learning in end-to-end SLT by presenting a multilingual approach to the task. Multilingual solutions are widely studied in MT and usually rely on “target forcing”, in which multilingual parallel data are combined to train a single model by prepending to the input sequences a language token that specifies the target language. However, when tested in speech translation, our experiments show that MT-like target forcing, used as is, is not effective in discriminating among the target languages. Thus, we propose a variant that uses target-language embed-dings to shift the input representations in different portions of the space according to the language, so to better support the production of output in the desired target language. Our experiments on end-to-end SLT from English into six languages show important improvements when translating into similar languages, especially when these are supported by scarce data. Further improvements are obtained when using English ASR data as an additional language (up to +2.5 BLEU points)." @default.
- W3008125272 created "2020-03-06" @default.
- W3008125272 creator A5015363344 @default.
- W3008125272 creator A5025827839 @default.
- W3008125272 creator A5048537874 @default.
- W3008125272 date "2019-12-01" @default.
- W3008125272 modified "2023-09-24" @default.
- W3008125272 title "One-to-Many Multilingual End-to-End Speech Translation" @default.
- W3008125272 cites W1836465849 @default.
- W3008125272 cites W2064675550 @default.
- W3008125272 cites W2112796928 @default.
- W3008125272 cites W2130942839 @default.
- W3008125272 cites W2148154194 @default.
- W3008125272 cites W2193413348 @default.
- W3008125272 cites W2194775991 @default.
- W3008125272 cites W2251743902 @default.
- W3008125272 cites W2327501763 @default.
- W3008125272 cites W2396144419 @default.
- W3008125272 cites W2550821151 @default.
- W3008125272 cites W2555745756 @default.
- W3008125272 cites W2582956876 @default.
- W3008125272 cites W2593011301 @default.
- W3008125272 cites W2605131327 @default.
- W3008125272 cites W2772120246 @default.
- W3008125272 cites W2794365787 @default.
- W3008125272 cites W2799800213 @default.
- W3008125272 cites W2808640845 @default.
- W3008125272 cites W2892009249 @default.
- W3008125272 cites W2899771611 @default.
- W3008125272 cites W2903193068 @default.
- W3008125272 cites W2936969148 @default.
- W3008125272 cites W2945700568 @default.
- W3008125272 cites W2962680099 @default.
- W3008125272 cites W2962778428 @default.
- W3008125272 cites W2962784628 @default.
- W3008125272 cites W2962824709 @default.
- W3008125272 cites W2963247703 @default.
- W3008125272 cites W2963403868 @default.
- W3008125272 cites W2963499433 @default.
- W3008125272 cites W2963506925 @default.
- W3008125272 cites W2963779652 @default.
- W3008125272 cites W2963842982 @default.
- W3008125272 cites W2963899033 @default.
- W3008125272 cites W2964007535 @default.
- W3008125272 cites W2964034111 @default.
- W3008125272 cites W2964104866 @default.
- W3008125272 cites W2964121744 @default.
- W3008125272 cites W2964161387 @default.
- W3008125272 cites W2964172053 @default.
- W3008125272 cites W2964265128 @default.
- W3008125272 cites W2964309797 @default.
- W3008125272 cites W2972448360 @default.
- W3008125272 cites W2973048981 @default.
- W3008125272 doi "https://doi.org/10.1109/asru46091.2019.9004003" @default.
- W3008125272 hasPublicationYear "2019" @default.
- W3008125272 type Work @default.
- W3008125272 sameAs 3008125272 @default.
- W3008125272 citedByCount "33" @default.
- W3008125272 countsByYear W30081252722020 @default.
- W3008125272 countsByYear W30081252722021 @default.
- W3008125272 countsByYear W30081252722022 @default.
- W3008125272 countsByYear W30081252722023 @default.
- W3008125272 crossrefType "proceedings-article" @default.
- W3008125272 hasAuthorship W3008125272A5015363344 @default.
- W3008125272 hasAuthorship W3008125272A5025827839 @default.
- W3008125272 hasAuthorship W3008125272A5048537874 @default.
- W3008125272 hasBestOaLocation W30081252722 @default.
- W3008125272 hasConcept C104317684 @default.
- W3008125272 hasConcept C105580179 @default.
- W3008125272 hasConcept C127313418 @default.
- W3008125272 hasConcept C137293760 @default.
- W3008125272 hasConcept C149364088 @default.
- W3008125272 hasConcept C154945302 @default.
- W3008125272 hasConcept C162324750 @default.
- W3008125272 hasConcept C185592680 @default.
- W3008125272 hasConcept C187736073 @default.
- W3008125272 hasConcept C197115733 @default.
- W3008125272 hasConcept C203005215 @default.
- W3008125272 hasConcept C204321447 @default.
- W3008125272 hasConcept C2780366754 @default.
- W3008125272 hasConcept C2780451532 @default.
- W3008125272 hasConcept C28490314 @default.
- W3008125272 hasConcept C41008148 @default.
- W3008125272 hasConcept C49204034 @default.
- W3008125272 hasConcept C55493867 @default.
- W3008125272 hasConcept C74296488 @default.
- W3008125272 hasConceptScore W3008125272C104317684 @default.
- W3008125272 hasConceptScore W3008125272C105580179 @default.
- W3008125272 hasConceptScore W3008125272C127313418 @default.
- W3008125272 hasConceptScore W3008125272C137293760 @default.
- W3008125272 hasConceptScore W3008125272C149364088 @default.
- W3008125272 hasConceptScore W3008125272C154945302 @default.
- W3008125272 hasConceptScore W3008125272C162324750 @default.
- W3008125272 hasConceptScore W3008125272C185592680 @default.
- W3008125272 hasConceptScore W3008125272C187736073 @default.
- W3008125272 hasConceptScore W3008125272C197115733 @default.
- W3008125272 hasConceptScore W3008125272C203005215 @default.
- W3008125272 hasConceptScore W3008125272C204321447 @default.
- W3008125272 hasConceptScore W3008125272C2780366754 @default.
- W3008125272 hasConceptScore W3008125272C2780451532 @default.