Matches in SemOpenAlex for { <https://semopenalex.org/work/W2970292446> ?p ?o ?g. }
- W2970292446 abstract "Learning target side syntactic structure has been shown to improve Neural Machine Translation (NMT). However, incorporating syntax through latent variables introduces additional complexity in inference, as the models need to marginalize over the latent syntactic structures. To avoid this, models often resort to greedy search which only allows them to explore a limited portion of the latent space. In this work, we introduce a new latent variable model, LaSyn, that captures the co-dependence between syntax and semantics, while allowing for effective and efficient inference over the latent space. LaSyn decouples direct dependence between successive latent variables, which allows its decoder to exhaustively search through the latent syntactic choices, while keeping decoding speed proportional to the size of the latent variable vocabulary. We implement LaSyn by modifying a transformer-based NMT system and design a neural expectation maximization algorithm that we regularize with part-of-speech information as the latent sequences. Evaluations on four different MT tasks show that incorporating target side syntax with LaSyn improves both translation quality, and also provides an opportunity to improve diversity." @default.
- W2970292446 created "2019-09-05" @default.
- W2970292446 creator A5006822602 @default.
- W2970292446 creator A5018054686 @default.
- W2970292446 creator A5036705554 @default.
- W2970292446 creator A5054023368 @default.
- W2970292446 creator A5054069581 @default.
- W2970292446 date "2019-01-01" @default.
- W2970292446 modified "2023-09-25" @default.
- W2970292446 title "Latent Part-of-Speech Sequences for Neural Machine Translation" @default.
- W2970292446 cites W1663973292 @default.
- W2970292446 cites W1902237438 @default.
- W2970292446 cites W2053742104 @default.
- W2970292446 cites W2101105183 @default.
- W2970292446 cites W2106751371 @default.
- W2970292446 cites W2140343992 @default.
- W2970292446 cites W2166905217 @default.
- W2970292446 cites W2222235228 @default.
- W2970292446 cites W2581624817 @default.
- W2970292446 cites W2594047108 @default.
- W2970292446 cites W2739894144 @default.
- W2970292446 cites W2785093437 @default.
- W2970292446 cites W2788277448 @default.
- W2970292446 cites W2798465082 @default.
- W2970292446 cites W2800490029 @default.
- W2970292446 cites W2803369080 @default.
- W2970292446 cites W2886026946 @default.
- W2970292446 cites W2889404673 @default.
- W2970292446 cites W2890908793 @default.
- W2970292446 cites W2890909908 @default.
- W2970292446 cites W2902441267 @default.
- W2970292446 cites W2907222274 @default.
- W2970292446 cites W2962717763 @default.
- W2970292446 cites W2962811598 @default.
- W2970292446 cites W2963011474 @default.
- W2970292446 cites W2963206148 @default.
- W2970292446 cites W2963216553 @default.
- W2970292446 cites W2963403868 @default.
- W2970292446 cites W2963648186 @default.
- W2970292446 cites W2963653811 @default.
- W2970292446 cites W2963661253 @default.
- W2970292446 cites W2963876447 @default.
- W2970292446 cites W2964048171 @default.
- W2970292446 cites W2964121744 @default.
- W2970292446 cites W2964265128 @default.
- W2970292446 cites W2964308564 @default.
- W2970292446 doi "https://doi.org/10.18653/v1/d19-1072" @default.
- W2970292446 hasPublicationYear "2019" @default.
- W2970292446 type Work @default.
- W2970292446 sameAs 2970292446 @default.
- W2970292446 citedByCount "6" @default.
- W2970292446 countsByYear W29702924462019 @default.
- W2970292446 countsByYear W29702924462020 @default.
- W2970292446 countsByYear W29702924462022 @default.
- W2970292446 crossrefType "proceedings-article" @default.
- W2970292446 hasAuthorship W2970292446A5006822602 @default.
- W2970292446 hasAuthorship W2970292446A5018054686 @default.
- W2970292446 hasAuthorship W2970292446A5036705554 @default.
- W2970292446 hasAuthorship W2970292446A5054023368 @default.
- W2970292446 hasAuthorship W2970292446A5054069581 @default.
- W2970292446 hasBestOaLocation W29702924462 @default.
- W2970292446 hasConcept C104317684 @default.
- W2970292446 hasConcept C105580179 @default.
- W2970292446 hasConcept C149364088 @default.
- W2970292446 hasConcept C154945302 @default.
- W2970292446 hasConcept C203005215 @default.
- W2970292446 hasConcept C204321447 @default.
- W2970292446 hasConcept C24687705 @default.
- W2970292446 hasConcept C2780366754 @default.
- W2970292446 hasConcept C28490314 @default.
- W2970292446 hasConcept C41008148 @default.
- W2970292446 hasConcept C55493867 @default.
- W2970292446 hasConcept C86803240 @default.
- W2970292446 hasConceptScore W2970292446C104317684 @default.
- W2970292446 hasConceptScore W2970292446C105580179 @default.
- W2970292446 hasConceptScore W2970292446C149364088 @default.
- W2970292446 hasConceptScore W2970292446C154945302 @default.
- W2970292446 hasConceptScore W2970292446C203005215 @default.
- W2970292446 hasConceptScore W2970292446C204321447 @default.
- W2970292446 hasConceptScore W2970292446C24687705 @default.
- W2970292446 hasConceptScore W2970292446C2780366754 @default.
- W2970292446 hasConceptScore W2970292446C28490314 @default.
- W2970292446 hasConceptScore W2970292446C41008148 @default.
- W2970292446 hasConceptScore W2970292446C55493867 @default.
- W2970292446 hasConceptScore W2970292446C86803240 @default.
- W2970292446 hasLocation W29702924461 @default.
- W2970292446 hasLocation W29702924462 @default.
- W2970292446 hasOpenAccess W2970292446 @default.
- W2970292446 hasPrimaryLocation W29702924461 @default.
- W2970292446 hasRelatedWork W1592339875 @default.
- W2970292446 hasRelatedWork W2010917994 @default.
- W2970292446 hasRelatedWork W2063433122 @default.
- W2970292446 hasRelatedWork W2128876910 @default.
- W2970292446 hasRelatedWork W2134352815 @default.
- W2970292446 hasRelatedWork W2134546430 @default.
- W2970292446 hasRelatedWork W2175555681 @default.
- W2970292446 hasRelatedWork W3107474891 @default.
- W2970292446 hasRelatedWork W626175318 @default.
- W2970292446 hasRelatedWork W2610387714 @default.
- W2970292446 isParatext "false" @default.