FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W3018753103> ?p ?o ?g. }

• W3018753103 abstract "Non-autoregressive (NAR) models generate all the tokens of a sequence in parallel, resulting in faster generation speed compared to their autoregressive (AR) counterparts but at the cost of lower accuracy. Different techniques including knowledge distillation and source-target alignment have been proposed to bridge the gap between AR and NAR models in various tasks such as neural machine translation (NMT), automatic speech recognition (ASR), and text to speech (TTS). With the help of those techniques, NAR models can catch up with the accuracy of AR models in some tasks but not in some others. In this work, we conduct a study to understand the difficulty of NAR sequence generation and try to answer: (1) Why NAR models can catch up with AR models in some tasks but not all? (2) Why techniques like knowledge distillation and source-target alignment can help NAR models. Since the main difference between AR and NAR models is that NAR models do not use dependency among target tokens while AR models do, intuitively the difficulty of NAR sequence generation heavily depends on the strongness of dependency among target tokens. To quantify such dependency, we propose an analysis model called CoMMA to characterize the difficulty of different NAR sequence generation tasks. We have several interesting findings: 1) Among the NMT, ASR and TTS tasks, ASR has the most target-token dependency while TTS has the least. 2) Knowledge distillation reduces the target-token dependency in target sequence and thus improves the accuracy of NAR models. 3) Source-target alignment constraint encourages dependency of a target token on source tokens and thus eases the training of NAR models." @default.
• W3018753103 created "2020-05-01" @default.
• W3018753103 creator A5018286848 @default.
• W3018753103 creator A5065126806 @default.
• W3018753103 creator A5070990160 @default.
• W3018753103 creator A5079260216 @default.
• W3018753103 creator A5088179161 @default.
• W3018753103 creator A5088930074 @default.
• W3018753103 date "2020-04-22" @default.
• W3018753103 modified "2023-10-16" @default.
• W3018753103 title "A Study of Non-autoregressive Model for Sequence Generation" @default.
• W3018753103 cites W1821462560 @default.
• W3018753103 cites W2592335154 @default.
• W3018753103 cites W2769810959 @default.
• W3018753103 cites W2787017828 @default.
• W3018753103 cites W2803985397 @default.
• W3018753103 cites W2890964657 @default.
• W3018753103 cites W2903739847 @default.
• W3018753103 cites W2905933322 @default.
• W3018753103 cites W2937297214 @default.
• W3018753103 cites W2945289329 @default.
• W3018753103 cites W2962784628 @default.
• W3018753103 cites W2962969034 @default.
• W3018753103 cites W2963300588 @default.
• W3018753103 cites W2963341956 @default.
• W3018753103 cites W2963403868 @default.
• W3018753103 cites W2963434219 @default.
• W3018753103 cites W2963536265 @default.
• W3018753103 cites W2964222566 @default.
• W3018753103 cites W2964243274 @default.
• W3018753103 cites W2970730223 @default.
• W3018753103 cites W2972448360 @default.
• W3018753103 cites W2972677740 @default.
• W3018753103 cites W2976135203 @default.
• W3018753103 cites W2976965654 @default.
• W3018753103 cites W2989134874 @default.
• W3018753103 cites W2996987694 @default.
• W3018753103 cites W3211259717 @default.
• W3018753103 doi "https://doi.org/10.48550/arxiv.2004.10454" @default.
• W3018753103 hasPublicationYear "2020" @default.
• W3018753103 type Work @default.
• W3018753103 sameAs 3018753103 @default.
• W3018753103 citedByCount "0" @default.
• W3018753103 crossrefType "posted-content" @default.
• W3018753103 hasAuthorship W3018753103A5018286848 @default.
• W3018753103 hasAuthorship W3018753103A5065126806 @default.
• W3018753103 hasAuthorship W3018753103A5070990160 @default.
• W3018753103 hasAuthorship W3018753103A5079260216 @default.
• W3018753103 hasAuthorship W3018753103A5088179161 @default.
• W3018753103 hasAuthorship W3018753103A5088930074 @default.
• W3018753103 hasBestOaLocation W30187531031 @default.
• W3018753103 hasConcept C105795698 @default.
• W3018753103 hasConcept C153180895 @default.
• W3018753103 hasConcept C154945302 @default.
• W3018753103 hasConcept C159877910 @default.
• W3018753103 hasConcept C19768560 @default.
• W3018753103 hasConcept C2524010 @default.
• W3018753103 hasConcept C2776036281 @default.
• W3018753103 hasConcept C2778112365 @default.
• W3018753103 hasConcept C28490314 @default.
• W3018753103 hasConcept C33923547 @default.
• W3018753103 hasConcept C38652104 @default.
• W3018753103 hasConcept C41008148 @default.
• W3018753103 hasConcept C48145219 @default.
• W3018753103 hasConcept C54355233 @default.
• W3018753103 hasConcept C86803240 @default.
• W3018753103 hasConceptScore W3018753103C105795698 @default.
• W3018753103 hasConceptScore W3018753103C153180895 @default.
• W3018753103 hasConceptScore W3018753103C154945302 @default.
• W3018753103 hasConceptScore W3018753103C159877910 @default.
• W3018753103 hasConceptScore W3018753103C19768560 @default.
• W3018753103 hasConceptScore W3018753103C2524010 @default.
• W3018753103 hasConceptScore W3018753103C2776036281 @default.
• W3018753103 hasConceptScore W3018753103C2778112365 @default.
• W3018753103 hasConceptScore W3018753103C28490314 @default.
• W3018753103 hasConceptScore W3018753103C33923547 @default.
• W3018753103 hasConceptScore W3018753103C38652104 @default.
• W3018753103 hasConceptScore W3018753103C41008148 @default.
• W3018753103 hasConceptScore W3018753103C48145219 @default.
• W3018753103 hasConceptScore W3018753103C54355233 @default.
• W3018753103 hasConceptScore W3018753103C86803240 @default.
• W3018753103 hasLocation W30187531031 @default.
• W3018753103 hasOpenAccess W3018753103 @default.
• W3018753103 hasPrimaryLocation W30187531031 @default.
• W3018753103 hasRelatedWork W1529400504 @default.
• W3018753103 hasRelatedWork W2123996664 @default.
• W3018753103 hasRelatedWork W2372328424 @default.
• W3018753103 hasRelatedWork W2375389409 @default.
• W3018753103 hasRelatedWork W2488051804 @default.
• W3018753103 hasRelatedWork W3018753103 @default.
• W3018753103 hasRelatedWork W3034363136 @default.
• W3018753103 hasRelatedWork W3183166027 @default.
• W3018753103 hasRelatedWork W3198116002 @default.
• W3018753103 hasRelatedWork W63071447 @default.
• W3018753103 isParatext "false" @default.
• W3018753103 isRetracted "false" @default.
• W3018753103 magId "3018753103" @default.
• W3018753103 workType "article" @default.