Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287204338> ?p ?o ?g. }
Showing items 1 to 87 of
87
with 100 items per page.
- W4287204338 abstract "Disfluency detection models now approach high accuracy on English text. However, little exploration has been done in improving the size and inference time of the model. At the same time, automatic speech recognition (ASR) models are moving from server-side inference to local, on-device inference. Supporting models in the transcription pipeline (like disfluency detection) must follow suit. In this work we concentrate on the disfluency detection task, focusing on small, fast, on-device models based on the BERT architecture. We demonstrate it is possible to train disfluency detection models as small as 1.3 MiB, while retaining high performance. We build on previous work that showed the benefit of data augmentation approaches such as self-training. Then, we evaluate the effect of domain mismatch between conversational and written text on model performance. We find that domain adaptation and data augmentation strategies have a more pronounced effect on these smaller models, as compared to conventional BERT models." @default.
- W4287204338 created "2022-07-25" @default.
- W4287204338 creator A5013219521 @default.
- W4287204338 creator A5028681799 @default.
- W4287204338 creator A5039601810 @default.
- W4287204338 creator A5055966977 @default.
- W4287204338 creator A5056130241 @default.
- W4287204338 creator A5059138163 @default.
- W4287204338 date "2021-04-21" @default.
- W4287204338 modified "2023-10-18" @default.
- W4287204338 title "Disfluency Detection with Unlabeled Data and Small BERT Models" @default.
- W4287204338 doi "https://doi.org/10.48550/arxiv.2104.10769" @default.
- W4287204338 hasPublicationYear "2021" @default.
- W4287204338 type Work @default.
- W4287204338 citedByCount "0" @default.
- W4287204338 crossrefType "posted-content" @default.
- W4287204338 hasAuthorship W4287204338A5013219521 @default.
- W4287204338 hasAuthorship W4287204338A5028681799 @default.
- W4287204338 hasAuthorship W4287204338A5039601810 @default.
- W4287204338 hasAuthorship W4287204338A5055966977 @default.
- W4287204338 hasAuthorship W4287204338A5056130241 @default.
- W4287204338 hasAuthorship W4287204338A5059138163 @default.
- W4287204338 hasBestOaLocation W42872043381 @default.
- W4287204338 hasConcept C119857082 @default.
- W4287204338 hasConcept C120665830 @default.
- W4287204338 hasConcept C121332964 @default.
- W4287204338 hasConcept C123657996 @default.
- W4287204338 hasConcept C134306372 @default.
- W4287204338 hasConcept C137293760 @default.
- W4287204338 hasConcept C139807058 @default.
- W4287204338 hasConcept C142362112 @default.
- W4287204338 hasConcept C153349607 @default.
- W4287204338 hasConcept C154945302 @default.
- W4287204338 hasConcept C162324750 @default.
- W4287204338 hasConcept C187736073 @default.
- W4287204338 hasConcept C199360897 @default.
- W4287204338 hasConcept C204321447 @default.
- W4287204338 hasConcept C2776145971 @default.
- W4287204338 hasConcept C2776214188 @default.
- W4287204338 hasConcept C2776434776 @default.
- W4287204338 hasConcept C2780451532 @default.
- W4287204338 hasConcept C28490314 @default.
- W4287204338 hasConcept C33923547 @default.
- W4287204338 hasConcept C36503486 @default.
- W4287204338 hasConcept C41008148 @default.
- W4287204338 hasConcept C43521106 @default.
- W4287204338 hasConcept C95623464 @default.
- W4287204338 hasConceptScore W4287204338C119857082 @default.
- W4287204338 hasConceptScore W4287204338C120665830 @default.
- W4287204338 hasConceptScore W4287204338C121332964 @default.
- W4287204338 hasConceptScore W4287204338C123657996 @default.
- W4287204338 hasConceptScore W4287204338C134306372 @default.
- W4287204338 hasConceptScore W4287204338C137293760 @default.
- W4287204338 hasConceptScore W4287204338C139807058 @default.
- W4287204338 hasConceptScore W4287204338C142362112 @default.
- W4287204338 hasConceptScore W4287204338C153349607 @default.
- W4287204338 hasConceptScore W4287204338C154945302 @default.
- W4287204338 hasConceptScore W4287204338C162324750 @default.
- W4287204338 hasConceptScore W4287204338C187736073 @default.
- W4287204338 hasConceptScore W4287204338C199360897 @default.
- W4287204338 hasConceptScore W4287204338C204321447 @default.
- W4287204338 hasConceptScore W4287204338C2776145971 @default.
- W4287204338 hasConceptScore W4287204338C2776214188 @default.
- W4287204338 hasConceptScore W4287204338C2776434776 @default.
- W4287204338 hasConceptScore W4287204338C2780451532 @default.
- W4287204338 hasConceptScore W4287204338C28490314 @default.
- W4287204338 hasConceptScore W4287204338C33923547 @default.
- W4287204338 hasConceptScore W4287204338C36503486 @default.
- W4287204338 hasConceptScore W4287204338C41008148 @default.
- W4287204338 hasConceptScore W4287204338C43521106 @default.
- W4287204338 hasConceptScore W4287204338C95623464 @default.
- W4287204338 hasLocation W42872043381 @default.
- W4287204338 hasOpenAccess W4287204338 @default.
- W4287204338 hasPrimaryLocation W42872043381 @default.
- W4287204338 hasRelatedWork W1839843306 @default.
- W4287204338 hasRelatedWork W2750946167 @default.
- W4287204338 hasRelatedWork W2963500702 @default.
- W4287204338 hasRelatedWork W3034238904 @default.
- W4287204338 hasRelatedWork W3107474891 @default.
- W4287204338 hasRelatedWork W3155411010 @default.
- W4287204338 hasRelatedWork W3185185774 @default.
- W4287204338 hasRelatedWork W3197433369 @default.
- W4287204338 hasRelatedWork W4287204338 @default.
- W4287204338 hasRelatedWork W4287812619 @default.
- W4287204338 isParatext "false" @default.
- W4287204338 isRetracted "false" @default.
- W4287204338 workType "article" @default.