Matches in SemOpenAlex for { <https://semopenalex.org/work/W3206096632> ?p ?o ?g. }
- W3206096632 abstract "Neural Machine Translation (NMT) models are known to suffer from noisy inputs. To make models robust, we generate adversarial augmentation samples that attack the model and preserve the source-side semantic meaning at the same time. To generate such samples, we propose a doubly-trained architecture that pairs two NMT models of opposite translation directions with a joint loss function, which combines the target-side attack and the source-side semantic similarity constraint. The results from our experiments across three different language pairs and two evaluation metrics show that these adversarial samples improve the model robustness." @default.
- W3206096632 created "2021-10-25" @default.
- W3206096632 creator A5010509940 @default.
- W3206096632 creator A5034588377 @default.
- W3206096632 creator A5080544452 @default.
- W3206096632 creator A5089260563 @default.
- W3206096632 date "2021-10-12" @default.
- W3206096632 modified "2023-09-28" @default.
- W3206096632 title "Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation." @default.
- W3206096632 cites W2101105183 @default.
- W3206096632 cites W2123301721 @default.
- W3206096632 cites W2777449390 @default.
- W3206096632 cites W2948099658 @default.
- W3206096632 cites W2950352427 @default.
- W3206096632 cites W2950359962 @default.
- W3206096632 cites W2950651087 @default.
- W3206096632 cites W2951467133 @default.
- W3206096632 cites W2952509486 @default.
- W3206096632 cites W2954157266 @default.
- W3206096632 cites W2963207607 @default.
- W3206096632 cites W2963463964 @default.
- W3206096632 cites W2963532001 @default.
- W3206096632 cites W2964247056 @default.
- W3206096632 cites W2970069595 @default.
- W3206096632 cites W2970250638 @default.
- W3206096632 cites W2971078337 @default.
- W3206096632 cites W2984051011 @default.
- W3206096632 cites W2996403597 @default.
- W3206096632 cites W3034397670 @default.
- W3206096632 cites W3035219095 @default.
- W3206096632 cites W3036558623 @default.
- W3206096632 cites W3096165964 @default.
- W3206096632 cites W3105214104 @default.
- W3206096632 cites W3146244013 @default.
- W3206096632 cites W3153818079 @default.
- W3206096632 hasPublicationYear "2021" @default.
- W3206096632 type Work @default.
- W3206096632 sameAs 3206096632 @default.
- W3206096632 citedByCount "0" @default.
- W3206096632 crossrefType "posted-content" @default.
- W3206096632 hasAuthorship W3206096632A5010509940 @default.
- W3206096632 hasAuthorship W3206096632A5034588377 @default.
- W3206096632 hasAuthorship W3206096632A5080544452 @default.
- W3206096632 hasAuthorship W3206096632A5089260563 @default.
- W3206096632 hasConcept C103278499 @default.
- W3206096632 hasConcept C104317684 @default.
- W3206096632 hasConcept C105580179 @default.
- W3206096632 hasConcept C115961682 @default.
- W3206096632 hasConcept C119857082 @default.
- W3206096632 hasConcept C130318100 @default.
- W3206096632 hasConcept C149364088 @default.
- W3206096632 hasConcept C154945302 @default.
- W3206096632 hasConcept C185592680 @default.
- W3206096632 hasConcept C203005215 @default.
- W3206096632 hasConcept C204321447 @default.
- W3206096632 hasConcept C2524010 @default.
- W3206096632 hasConcept C2776036281 @default.
- W3206096632 hasConcept C2984842247 @default.
- W3206096632 hasConcept C33923547 @default.
- W3206096632 hasConcept C37736160 @default.
- W3206096632 hasConcept C41008148 @default.
- W3206096632 hasConcept C50644808 @default.
- W3206096632 hasConcept C55493867 @default.
- W3206096632 hasConcept C63479239 @default.
- W3206096632 hasConceptScore W3206096632C103278499 @default.
- W3206096632 hasConceptScore W3206096632C104317684 @default.
- W3206096632 hasConceptScore W3206096632C105580179 @default.
- W3206096632 hasConceptScore W3206096632C115961682 @default.
- W3206096632 hasConceptScore W3206096632C119857082 @default.
- W3206096632 hasConceptScore W3206096632C130318100 @default.
- W3206096632 hasConceptScore W3206096632C149364088 @default.
- W3206096632 hasConceptScore W3206096632C154945302 @default.
- W3206096632 hasConceptScore W3206096632C185592680 @default.
- W3206096632 hasConceptScore W3206096632C203005215 @default.
- W3206096632 hasConceptScore W3206096632C204321447 @default.
- W3206096632 hasConceptScore W3206096632C2524010 @default.
- W3206096632 hasConceptScore W3206096632C2776036281 @default.
- W3206096632 hasConceptScore W3206096632C2984842247 @default.
- W3206096632 hasConceptScore W3206096632C33923547 @default.
- W3206096632 hasConceptScore W3206096632C37736160 @default.
- W3206096632 hasConceptScore W3206096632C41008148 @default.
- W3206096632 hasConceptScore W3206096632C50644808 @default.
- W3206096632 hasConceptScore W3206096632C55493867 @default.
- W3206096632 hasConceptScore W3206096632C63479239 @default.
- W3206096632 hasOpenAccess W3206096632 @default.
- W3206096632 hasRelatedWork W1577269164 @default.
- W3206096632 hasRelatedWork W2493519722 @default.
- W3206096632 hasRelatedWork W2781800156 @default.
- W3206096632 hasRelatedWork W2798474427 @default.
- W3206096632 hasRelatedWork W2799012717 @default.
- W3206096632 hasRelatedWork W2891524928 @default.
- W3206096632 hasRelatedWork W2899135636 @default.
- W3206096632 hasRelatedWork W2902792509 @default.
- W3206096632 hasRelatedWork W2950848235 @default.
- W3206096632 hasRelatedWork W2951295768 @default.
- W3206096632 hasRelatedWork W2966269638 @default.
- W3206096632 hasRelatedWork W2969026700 @default.
- W3206096632 hasRelatedWork W2970065861 @default.
- W3206096632 hasRelatedWork W2970694516 @default.
- W3206096632 hasRelatedWork W2977418027 @default.