Matches in SemOpenAlex for { <https://semopenalex.org/work/W2798400986> ?p ?o ?g. }
- W2798400986 abstract "Neural machine translation (NMT) models are typically trained with fixed-size input and output vocabularies, which creates an important bottleneck on their accuracy and generalization capability. As a solution, various studies proposed segmenting words into sub-word units and performing translation at the sub-lexical level. However, statistical word segmentation methods have recently shown to be prone to morphological errors, which can lead to inaccurate translations. In this paper, we propose to overcome this problem by replacing the source-language embedding layer of NMT with a bi-directional recurrent neural network that generates compositional representations of the input at any desired level of granularity. We test our approach in a low-resource setting with five languages from different morphological typologies, and under different composition assumptions. By training NMT to compose word representations from character n-grams, our approach consistently outperforms (from 1.71 to 2.48 BLEU points) NMT learning embeddings of statistically generated sub-word units." @default.
- W2798400986 created "2018-05-07" @default.
- W2798400986 creator A5068340972 @default.
- W2798400986 creator A5086132343 @default.
- W2798400986 date "2018-05-05" @default.
- W2798400986 modified "2023-09-27" @default.
- W2798400986 title "Compositional Representation of Morphologically-Rich Input for Neural Machine Translation" @default.
- W2798400986 cites W114517082 @default.
- W2798400986 cites W1957775505 @default.
- W2798400986 cites W2064675550 @default.
- W2798400986 cites W2101105183 @default.
- W2798400986 cites W2144600658 @default.
- W2798400986 cites W2146502635 @default.
- W2798400986 cites W2150355110 @default.
- W2798400986 cites W2172140247 @default.
- W2798400986 cites W2184135559 @default.
- W2798400986 cites W2220350356 @default.
- W2798400986 cites W2384495648 @default.
- W2798400986 cites W2525778437 @default.
- W2798400986 cites W2531207078 @default.
- W2798400986 cites W2532807140 @default.
- W2798400986 cites W2594229957 @default.
- W2798400986 cites W2611767671 @default.
- W2798400986 cites W2621587759 @default.
- W2798400986 cites W2757041753 @default.
- W2798400986 cites W2886744048 @default.
- W2798400986 cites W2962784628 @default.
- W2798400986 cites W2963324947 @default.
- W2798400986 cites W2964308564 @default.
- W2798400986 cites W3005389111 @default.
- W2798400986 cites W46679369 @default.
- W2798400986 hasPublicationYear "2018" @default.
- W2798400986 type Work @default.
- W2798400986 sameAs 2798400986 @default.
- W2798400986 citedByCount "0" @default.
- W2798400986 crossrefType "posted-content" @default.
- W2798400986 hasAuthorship W2798400986A5068340972 @default.
- W2798400986 hasAuthorship W2798400986A5086132343 @default.
- W2798400986 hasConcept C104317684 @default.
- W2798400986 hasConcept C105580179 @default.
- W2798400986 hasConcept C111919701 @default.
- W2798400986 hasConcept C130597682 @default.
- W2798400986 hasConcept C134306372 @default.
- W2798400986 hasConcept C149364088 @default.
- W2798400986 hasConcept C149635348 @default.
- W2798400986 hasConcept C154945302 @default.
- W2798400986 hasConcept C177148314 @default.
- W2798400986 hasConcept C17744445 @default.
- W2798400986 hasConcept C177774035 @default.
- W2798400986 hasConcept C185592680 @default.
- W2798400986 hasConcept C199539241 @default.
- W2798400986 hasConcept C203005215 @default.
- W2798400986 hasConcept C204321447 @default.
- W2798400986 hasConcept C24687705 @default.
- W2798400986 hasConcept C2524010 @default.
- W2798400986 hasConcept C2776359362 @default.
- W2798400986 hasConcept C2777462759 @default.
- W2798400986 hasConcept C2780513914 @default.
- W2798400986 hasConcept C2780861071 @default.
- W2798400986 hasConcept C33923547 @default.
- W2798400986 hasConcept C41008148 @default.
- W2798400986 hasConcept C41608201 @default.
- W2798400986 hasConcept C50644808 @default.
- W2798400986 hasConcept C55493867 @default.
- W2798400986 hasConcept C90805587 @default.
- W2798400986 hasConcept C94625758 @default.
- W2798400986 hasConceptScore W2798400986C104317684 @default.
- W2798400986 hasConceptScore W2798400986C105580179 @default.
- W2798400986 hasConceptScore W2798400986C111919701 @default.
- W2798400986 hasConceptScore W2798400986C130597682 @default.
- W2798400986 hasConceptScore W2798400986C134306372 @default.
- W2798400986 hasConceptScore W2798400986C149364088 @default.
- W2798400986 hasConceptScore W2798400986C149635348 @default.
- W2798400986 hasConceptScore W2798400986C154945302 @default.
- W2798400986 hasConceptScore W2798400986C177148314 @default.
- W2798400986 hasConceptScore W2798400986C17744445 @default.
- W2798400986 hasConceptScore W2798400986C177774035 @default.
- W2798400986 hasConceptScore W2798400986C185592680 @default.
- W2798400986 hasConceptScore W2798400986C199539241 @default.
- W2798400986 hasConceptScore W2798400986C203005215 @default.
- W2798400986 hasConceptScore W2798400986C204321447 @default.
- W2798400986 hasConceptScore W2798400986C24687705 @default.
- W2798400986 hasConceptScore W2798400986C2524010 @default.
- W2798400986 hasConceptScore W2798400986C2776359362 @default.
- W2798400986 hasConceptScore W2798400986C2777462759 @default.
- W2798400986 hasConceptScore W2798400986C2780513914 @default.
- W2798400986 hasConceptScore W2798400986C2780861071 @default.
- W2798400986 hasConceptScore W2798400986C33923547 @default.
- W2798400986 hasConceptScore W2798400986C41008148 @default.
- W2798400986 hasConceptScore W2798400986C41608201 @default.
- W2798400986 hasConceptScore W2798400986C50644808 @default.
- W2798400986 hasConceptScore W2798400986C55493867 @default.
- W2798400986 hasConceptScore W2798400986C90805587 @default.
- W2798400986 hasConceptScore W2798400986C94625758 @default.
- W2798400986 hasLocation W27984009861 @default.
- W2798400986 hasOpenAccess W2798400986 @default.
- W2798400986 hasPrimaryLocation W27984009861 @default.
- W2798400986 hasRelatedWork W149848483 @default.
- W2798400986 hasRelatedWork W2410217169 @default.
- W2798400986 hasRelatedWork W2610661645 @default.