Matches in SemOpenAlex for { <https://semopenalex.org/work/W4224952741> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W4224952741 abstract "Transformer-based automatic speech recognition (ASR) systems have shown their success in the presence of large datasets. But, in medical research, we have to create ASR for the non-typical population, i.e. pre-school children with speech disorders, with small training dataset. To increase training efficiency on small datasets, we optimize the architecture of Wav2Vec 2.0, a variation of Transformer, through analyzing its pre-trained model’s block-level attention pattern. We show that block-level patterns can serve as an indicator for narrowing down the optimization direction. To ensure the reproducibility of our experiments, we leverage Librispeech-100-clean as training data to simulate the limited data condition. We leverage two techniques, local attention mechanism and cross-block parameter sharing, with counter-intuitive configurations. Our optimized architecture outperforms the vanilla architecture about 1.8% absolute word error rate (WER) on dev-clean and 1.4% on test-clean." @default.
- W4224952741 created "2022-04-28" @default.
- W4224952741 creator A5021979900 @default.
- W4224952741 creator A5038300369 @default.
- W4224952741 creator A5059905063 @default.
- W4224952741 date "2022-05-23" @default.
- W4224952741 modified "2023-10-16" @default.
- W4224952741 title "Optimize Wav2vec2s Architecture for Small Training Set Through Analyzing its Pre-Trained Models Attention Pattern" @default.
- W4224952741 cites W1494198834 @default.
- W4224952741 cites W2012710017 @default.
- W4224952741 cites W2127141656 @default.
- W4224952741 cites W2164089264 @default.
- W4224952741 cites W2933138175 @default.
- W4224952741 cites W2964110616 @default.
- W4224952741 cites W2970120757 @default.
- W4224952741 cites W3034561418 @default.
- W4224952741 cites W3097777922 @default.
- W4224952741 cites W3159719190 @default.
- W4224952741 doi "https://doi.org/10.1109/icassp43922.2022.9747831" @default.
- W4224952741 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37193061" @default.
- W4224952741 hasPublicationYear "2022" @default.
- W4224952741 type Work @default.
- W4224952741 citedByCount "0" @default.
- W4224952741 crossrefType "proceedings-article" @default.
- W4224952741 hasAuthorship W4224952741A5021979900 @default.
- W4224952741 hasAuthorship W4224952741A5038300369 @default.
- W4224952741 hasAuthorship W4224952741A5059905063 @default.
- W4224952741 hasConcept C119599485 @default.
- W4224952741 hasConcept C119857082 @default.
- W4224952741 hasConcept C123657996 @default.
- W4224952741 hasConcept C127413603 @default.
- W4224952741 hasConcept C142362112 @default.
- W4224952741 hasConcept C144024400 @default.
- W4224952741 hasConcept C149923435 @default.
- W4224952741 hasConcept C153083717 @default.
- W4224952741 hasConcept C153180895 @default.
- W4224952741 hasConcept C153349607 @default.
- W4224952741 hasConcept C154945302 @default.
- W4224952741 hasConcept C165801399 @default.
- W4224952741 hasConcept C28490314 @default.
- W4224952741 hasConcept C2908647359 @default.
- W4224952741 hasConcept C40969351 @default.
- W4224952741 hasConcept C41008148 @default.
- W4224952741 hasConcept C51632099 @default.
- W4224952741 hasConcept C66322947 @default.
- W4224952741 hasConceptScore W4224952741C119599485 @default.
- W4224952741 hasConceptScore W4224952741C119857082 @default.
- W4224952741 hasConceptScore W4224952741C123657996 @default.
- W4224952741 hasConceptScore W4224952741C127413603 @default.
- W4224952741 hasConceptScore W4224952741C142362112 @default.
- W4224952741 hasConceptScore W4224952741C144024400 @default.
- W4224952741 hasConceptScore W4224952741C149923435 @default.
- W4224952741 hasConceptScore W4224952741C153083717 @default.
- W4224952741 hasConceptScore W4224952741C153180895 @default.
- W4224952741 hasConceptScore W4224952741C153349607 @default.
- W4224952741 hasConceptScore W4224952741C154945302 @default.
- W4224952741 hasConceptScore W4224952741C165801399 @default.
- W4224952741 hasConceptScore W4224952741C28490314 @default.
- W4224952741 hasConceptScore W4224952741C2908647359 @default.
- W4224952741 hasConceptScore W4224952741C40969351 @default.
- W4224952741 hasConceptScore W4224952741C41008148 @default.
- W4224952741 hasConceptScore W4224952741C51632099 @default.
- W4224952741 hasConceptScore W4224952741C66322947 @default.
- W4224952741 hasLocation W42249527411 @default.
- W4224952741 hasLocation W42249527412 @default.
- W4224952741 hasOpenAccess W4224952741 @default.
- W4224952741 hasPrimaryLocation W42249527411 @default.
- W4224952741 hasRelatedWork W1965249131 @default.
- W4224952741 hasRelatedWork W2150750161 @default.
- W4224952741 hasRelatedWork W2285081780 @default.
- W4224952741 hasRelatedWork W2793122029 @default.
- W4224952741 hasRelatedWork W3030610722 @default.
- W4224952741 hasRelatedWork W3047941943 @default.
- W4224952741 hasRelatedWork W3095123370 @default.
- W4224952741 hasRelatedWork W4224239264 @default.
- W4224952741 hasRelatedWork W4294299598 @default.
- W4224952741 hasRelatedWork W4313440285 @default.
- W4224952741 isParatext "false" @default.
- W4224952741 isRetracted "false" @default.
- W4224952741 workType "article" @default.