Matches in SemOpenAlex for { <https://semopenalex.org/work/W3109234242> ?p ?o ?g. }
- W3109234242 abstract "The performance of automatic speech recognition (ASR) systems typically degrades significantly when the training and test data domains are mismatched. In this paper, we show that self-training (ST) combined with an uncertainty-based pseudo-label filtering approach can be effectively used for domain adaptation. We propose DUST, a dropout-based uncertainty-driven self-training technique which uses agreement between multiple predictions of an ASR system obtained for different dropout settings to measure the model's uncertainty about its prediction. DUST excludes pseudo-labeled data with high uncertainties from the training, which leads to substantially improved ASR results compared to ST without filtering, and accelerates the training time due to a reduced training data set. Domain adaptation experiments using WSJ as a source domain and TED-LIUM 3 as well as SWITCHBOARD as the target domains show that up to 80% of the performance of a system trained on ground-truth data can be recovered." @default.
- W3109234242 created "2020-12-07" @default.
- W3109234242 creator A5008673783 @default.
- W3109234242 creator A5071286568 @default.
- W3109234242 creator A5076453358 @default.
- W3109234242 creator A5087554069 @default.
- W3109234242 date "2020-11-26" @default.
- W3109234242 modified "2023-09-25" @default.
- W3109234242 title "Unsupervised Domain Adaptation for Speech Recognition via Uncertainty Driven Self-Training" @default.
- W3109234242 cites W1494198834 @default.
- W3109234242 cites W1585385982 @default.
- W3109234242 cites W1594039573 @default.
- W3109234242 cites W2024490156 @default.
- W3109234242 cites W2095705004 @default.
- W3109234242 cites W2097089247 @default.
- W3109234242 cites W2127141656 @default.
- W3109234242 cites W2133556223 @default.
- W3109234242 cites W2134800885 @default.
- W3109234242 cites W2143612262 @default.
- W3109234242 cites W2514741789 @default.
- W3109234242 cites W2696967604 @default.
- W3109234242 cites W2739883972 @default.
- W3109234242 cites W2892009249 @default.
- W3109234242 cites W2936774411 @default.
- W3109234242 cites W2939216642 @default.
- W3109234242 cites W2962684181 @default.
- W3109234242 cites W2962687275 @default.
- W3109234242 cites W2963431393 @default.
- W3109234242 cites W2964059111 @default.
- W3109234242 cites W2998532468 @default.
- W3109234242 cites W3027083471 @default.
- W3109234242 cites W3030437843 @default.
- W3109234242 cites W3049397270 @default.
- W3109234242 cites W3096273170 @default.
- W3109234242 cites W3099782249 @default.
- W3109234242 cites W3101648800 @default.
- W3109234242 cites W3162833755 @default.
- W3109234242 cites W854541894 @default.
- W3109234242 hasPublicationYear "2020" @default.
- W3109234242 type Work @default.
- W3109234242 sameAs 3109234242 @default.
- W3109234242 citedByCount "0" @default.
- W3109234242 crossrefType "posted-content" @default.
- W3109234242 hasAuthorship W3109234242A5008673783 @default.
- W3109234242 hasAuthorship W3109234242A5071286568 @default.
- W3109234242 hasAuthorship W3109234242A5076453358 @default.
- W3109234242 hasAuthorship W3109234242A5087554069 @default.
- W3109234242 hasConcept C119857082 @default.
- W3109234242 hasConcept C120665830 @default.
- W3109234242 hasConcept C121332964 @default.
- W3109234242 hasConcept C134306372 @default.
- W3109234242 hasConcept C139807058 @default.
- W3109234242 hasConcept C153180895 @default.
- W3109234242 hasConcept C153294291 @default.
- W3109234242 hasConcept C154945302 @default.
- W3109234242 hasConcept C177264268 @default.
- W3109234242 hasConcept C199360897 @default.
- W3109234242 hasConcept C2776145597 @default.
- W3109234242 hasConcept C2776434776 @default.
- W3109234242 hasConcept C2777211547 @default.
- W3109234242 hasConcept C28490314 @default.
- W3109234242 hasConcept C33923547 @default.
- W3109234242 hasConcept C36503486 @default.
- W3109234242 hasConcept C41008148 @default.
- W3109234242 hasConcept C51632099 @default.
- W3109234242 hasConcept C58489278 @default.
- W3109234242 hasConcept C95623464 @default.
- W3109234242 hasConceptScore W3109234242C119857082 @default.
- W3109234242 hasConceptScore W3109234242C120665830 @default.
- W3109234242 hasConceptScore W3109234242C121332964 @default.
- W3109234242 hasConceptScore W3109234242C134306372 @default.
- W3109234242 hasConceptScore W3109234242C139807058 @default.
- W3109234242 hasConceptScore W3109234242C153180895 @default.
- W3109234242 hasConceptScore W3109234242C153294291 @default.
- W3109234242 hasConceptScore W3109234242C154945302 @default.
- W3109234242 hasConceptScore W3109234242C177264268 @default.
- W3109234242 hasConceptScore W3109234242C199360897 @default.
- W3109234242 hasConceptScore W3109234242C2776145597 @default.
- W3109234242 hasConceptScore W3109234242C2776434776 @default.
- W3109234242 hasConceptScore W3109234242C2777211547 @default.
- W3109234242 hasConceptScore W3109234242C28490314 @default.
- W3109234242 hasConceptScore W3109234242C33923547 @default.
- W3109234242 hasConceptScore W3109234242C36503486 @default.
- W3109234242 hasConceptScore W3109234242C41008148 @default.
- W3109234242 hasConceptScore W3109234242C51632099 @default.
- W3109234242 hasConceptScore W3109234242C58489278 @default.
- W3109234242 hasConceptScore W3109234242C95623464 @default.
- W3109234242 hasLocation W31092342421 @default.
- W3109234242 hasOpenAccess W3109234242 @default.
- W3109234242 hasPrimaryLocation W31092342421 @default.
- W3109234242 hasRelatedWork W2548256801 @default.
- W3109234242 hasRelatedWork W2683320792 @default.
- W3109234242 hasRelatedWork W2737108017 @default.
- W3109234242 hasRelatedWork W2749784707 @default.
- W3109234242 hasRelatedWork W2766572290 @default.
- W3109234242 hasRelatedWork W2796010067 @default.
- W3109234242 hasRelatedWork W2890351244 @default.
- W3109234242 hasRelatedWork W2963122170 @default.
- W3109234242 hasRelatedWork W2964245029 @default.
- W3109234242 hasRelatedWork W2972981541 @default.