Matches in SemOpenAlex for { <https://semopenalex.org/work/W3167515372> ?p ?o ?g. }
- W3167515372 abstract "While it has been shown that Neural Machine Translation (NMT) is highly sensitive to noisy parallel training samples, prior work treats all types of mismatches between source and target as noise. As a result, it remains unclear how samples that are mostly equivalent but contain a small number of semantically divergent tokens impact NMT training. To close this gap, we analyze the impact of different types of fine-grained semantic divergences on Transformer models. We show that models trained on synthetic divergences output degenerated text more frequently and are less confident in their predictions. Based on these findings, we introduce a divergent-aware NMT framework that uses factors to help NMT recover from the degradation caused by naturally occurring divergences, improving both translation quality and model calibration on EN-FR tasks." @default.
- W3167515372 created "2021-06-22" @default.
- W3167515372 creator A5059930218 @default.
- W3167515372 creator A5078390032 @default.
- W3167515372 date "2021-05-31" @default.
- W3167515372 modified "2023-09-27" @default.
- W3167515372 title "Beyond Noise: Mitigating the Impact of Fine-grained Semantic Divergences on Neural Machine Translation" @default.
- W3167515372 cites W1988012253 @default.
- W3167515372 cites W2101105183 @default.
- W3167515372 cites W2105673178 @default.
- W3167515372 cites W2107695330 @default.
- W3167515372 cites W2123301721 @default.
- W3167515372 cites W2170716095 @default.
- W3167515372 cites W2467834614 @default.
- W3167515372 cites W2573665256 @default.
- W3167515372 cites W2741900445 @default.
- W3167515372 cites W2786991584 @default.
- W3167515372 cites W2790319220 @default.
- W3167515372 cites W2885950361 @default.
- W3167515372 cites W2902918014 @default.
- W3167515372 cites W2960374072 @default.
- W3167515372 cites W2962784628 @default.
- W3167515372 cites W2963216553 @default.
- W3167515372 cites W2963250244 @default.
- W3167515372 cites W2963919854 @default.
- W3167515372 cites W2964022663 @default.
- W3167515372 cites W2964121744 @default.
- W3167515372 cites W2964212410 @default.
- W3167515372 cites W2970858854 @default.
- W3167515372 cites W2972780142 @default.
- W3167515372 cites W2973088264 @default.
- W3167515372 cites W3022579377 @default.
- W3167515372 cites W3088421316 @default.
- W3167515372 cites W3098396250 @default.
- W3167515372 cites W3119872155 @default.
- W3167515372 cites W3171241877 @default.
- W3167515372 hasPublicationYear "2021" @default.
- W3167515372 type Work @default.
- W3167515372 sameAs 3167515372 @default.
- W3167515372 citedByCount "1" @default.
- W3167515372 countsByYear W31675153722021 @default.
- W3167515372 crossrefType "posted-content" @default.
- W3167515372 hasAuthorship W3167515372A5059930218 @default.
- W3167515372 hasAuthorship W3167515372A5078390032 @default.
- W3167515372 hasConcept C104317684 @default.
- W3167515372 hasConcept C105580179 @default.
- W3167515372 hasConcept C115961682 @default.
- W3167515372 hasConcept C119857082 @default.
- W3167515372 hasConcept C121332964 @default.
- W3167515372 hasConcept C149364088 @default.
- W3167515372 hasConcept C154945302 @default.
- W3167515372 hasConcept C165801399 @default.
- W3167515372 hasConcept C185592680 @default.
- W3167515372 hasConcept C203005215 @default.
- W3167515372 hasConcept C204321447 @default.
- W3167515372 hasConcept C2779530757 @default.
- W3167515372 hasConcept C41008148 @default.
- W3167515372 hasConcept C55493867 @default.
- W3167515372 hasConcept C62520636 @default.
- W3167515372 hasConcept C66322947 @default.
- W3167515372 hasConcept C99498987 @default.
- W3167515372 hasConceptScore W3167515372C104317684 @default.
- W3167515372 hasConceptScore W3167515372C105580179 @default.
- W3167515372 hasConceptScore W3167515372C115961682 @default.
- W3167515372 hasConceptScore W3167515372C119857082 @default.
- W3167515372 hasConceptScore W3167515372C121332964 @default.
- W3167515372 hasConceptScore W3167515372C149364088 @default.
- W3167515372 hasConceptScore W3167515372C154945302 @default.
- W3167515372 hasConceptScore W3167515372C165801399 @default.
- W3167515372 hasConceptScore W3167515372C185592680 @default.
- W3167515372 hasConceptScore W3167515372C203005215 @default.
- W3167515372 hasConceptScore W3167515372C204321447 @default.
- W3167515372 hasConceptScore W3167515372C2779530757 @default.
- W3167515372 hasConceptScore W3167515372C41008148 @default.
- W3167515372 hasConceptScore W3167515372C55493867 @default.
- W3167515372 hasConceptScore W3167515372C62520636 @default.
- W3167515372 hasConceptScore W3167515372C66322947 @default.
- W3167515372 hasConceptScore W3167515372C99498987 @default.
- W3167515372 hasLocation W31675153721 @default.
- W3167515372 hasOpenAccess W3167515372 @default.
- W3167515372 hasPrimaryLocation W31675153721 @default.
- W3167515372 hasRelatedWork W2576638397 @default.
- W3167515372 hasRelatedWork W2725170098 @default.
- W3167515372 hasRelatedWork W2755327023 @default.
- W3167515372 hasRelatedWork W2794607286 @default.
- W3167515372 hasRelatedWork W2946817437 @default.
- W3167515372 hasRelatedWork W2950118387 @default.
- W3167515372 hasRelatedWork W2982644924 @default.
- W3167515372 hasRelatedWork W2989208022 @default.
- W3167515372 hasRelatedWork W3007937028 @default.
- W3167515372 hasRelatedWork W3028787912 @default.
- W3167515372 hasRelatedWork W3091836629 @default.
- W3167515372 hasRelatedWork W3096237101 @default.
- W3167515372 hasRelatedWork W3173651177 @default.
- W3167515372 hasRelatedWork W3173710818 @default.
- W3167515372 hasRelatedWork W3176108833 @default.
- W3167515372 hasRelatedWork W3200765429 @default.
- W3167515372 hasRelatedWork W3201476335 @default.
- W3167515372 hasRelatedWork W3204051225 @default.
- W3167515372 hasRelatedWork W3208214534 @default.