Matches in SemOpenAlex for { <https://semopenalex.org/work/W1770758908> ?p ?o ?g. }
Showing items 1 to 90 of
90
with 100 items per page.
- W1770758908 abstract "To develop speaker adaptation algorithms for deep neural network (DNN) that are suitable for large-scale online deployment, it is desirable that the adaptation model be represented in a compact form and learned in an unsupervised fashion. In this paper, we propose a novel low-footprint adaptation technique for DNN that adapts the DNN model through node activation functions. The approach introduces slope and bias parameters in the sigmoid activation functions for each speaker, allowing the adaptation model to be stored in a small-sized storage space. We show that this adaptation technique can be formulated in a linear regression fashion, analogous to other speak adaptation algorithms that apply additional linear transformations to the DNN layers. We further investigate semi-supervised online adaptation by making use of the user click-through data as a supervision signal. The proposed method is evaluated on short message dictation and voice search tasks in supervised, unsupervised, and semi-supervised setups. Compared with the singular value decomposition (SVD) bottleneck adaptation, the proposed adaptation method achieves comparable accuracy improvements with much smaller footprint." @default.
- W1770758908 created "2016-06-24" @default.
- W1770758908 creator A5029670581 @default.
- W1770758908 creator A5049930476 @default.
- W1770758908 creator A5077401426 @default.
- W1770758908 creator A5080344490 @default.
- W1770758908 date "2015-04-01" @default.
- W1770758908 modified "2023-10-18" @default.
- W1770758908 title "Investigating online low-footprint speaker adaptation using generalized linear regression and click-through data" @default.
- W1770758908 cites W1513862252 @default.
- W1770758908 cites W1537275613 @default.
- W1770758908 cites W1984541135 @default.
- W1770758908 cites W1985371235 @default.
- W1770758908 cites W1989549063 @default.
- W1770758908 cites W2002342963 @default.
- W1770758908 cites W2036351241 @default.
- W1770758908 cites W2056738732 @default.
- W1770758908 cites W2076794394 @default.
- W1770758908 cites W2080005694 @default.
- W1770758908 cites W2087006792 @default.
- W1770758908 cites W2094147890 @default.
- W1770758908 cites W2095098834 @default.
- W1770758908 cites W2117239706 @default.
- W1770758908 cites W2124558353 @default.
- W1770758908 cites W2127499922 @default.
- W1770758908 cites W2143577772 @default.
- W1770758908 cites W2146871184 @default.
- W1770758908 cites W2147768505 @default.
- W1770758908 cites W2160306971 @default.
- W1770758908 cites W2173880944 @default.
- W1770758908 cites W2296748324 @default.
- W1770758908 cites W2397860787 @default.
- W1770758908 cites W82936479 @default.
- W1770758908 doi "https://doi.org/10.1109/icassp.2015.7178784" @default.
- W1770758908 hasPublicationYear "2015" @default.
- W1770758908 type Work @default.
- W1770758908 sameAs 1770758908 @default.
- W1770758908 citedByCount "30" @default.
- W1770758908 countsByYear W17707589082015 @default.
- W1770758908 countsByYear W17707589082016 @default.
- W1770758908 countsByYear W17707589082017 @default.
- W1770758908 countsByYear W17707589082018 @default.
- W1770758908 countsByYear W17707589082019 @default.
- W1770758908 countsByYear W17707589082020 @default.
- W1770758908 crossrefType "proceedings-article" @default.
- W1770758908 hasAuthorship W1770758908A5029670581 @default.
- W1770758908 hasAuthorship W1770758908A5049930476 @default.
- W1770758908 hasAuthorship W1770758908A5077401426 @default.
- W1770758908 hasAuthorship W1770758908A5080344490 @default.
- W1770758908 hasConcept C119857082 @default.
- W1770758908 hasConcept C120665830 @default.
- W1770758908 hasConcept C121332964 @default.
- W1770758908 hasConcept C139807058 @default.
- W1770758908 hasConcept C149635348 @default.
- W1770758908 hasConcept C153180895 @default.
- W1770758908 hasConcept C154945302 @default.
- W1770758908 hasConcept C22789450 @default.
- W1770758908 hasConcept C2780513914 @default.
- W1770758908 hasConcept C28490314 @default.
- W1770758908 hasConcept C41008148 @default.
- W1770758908 hasConcept C50644808 @default.
- W1770758908 hasConceptScore W1770758908C119857082 @default.
- W1770758908 hasConceptScore W1770758908C120665830 @default.
- W1770758908 hasConceptScore W1770758908C121332964 @default.
- W1770758908 hasConceptScore W1770758908C139807058 @default.
- W1770758908 hasConceptScore W1770758908C149635348 @default.
- W1770758908 hasConceptScore W1770758908C153180895 @default.
- W1770758908 hasConceptScore W1770758908C154945302 @default.
- W1770758908 hasConceptScore W1770758908C22789450 @default.
- W1770758908 hasConceptScore W1770758908C2780513914 @default.
- W1770758908 hasConceptScore W1770758908C28490314 @default.
- W1770758908 hasConceptScore W1770758908C41008148 @default.
- W1770758908 hasConceptScore W1770758908C50644808 @default.
- W1770758908 hasLocation W17707589081 @default.
- W1770758908 hasOpenAccess W1770758908 @default.
- W1770758908 hasPrimaryLocation W17707589081 @default.
- W1770758908 hasRelatedWork W2080597906 @default.
- W1770758908 hasRelatedWork W2087937280 @default.
- W1770758908 hasRelatedWork W2353647904 @default.
- W1770758908 hasRelatedWork W2354251581 @default.
- W1770758908 hasRelatedWork W2357461155 @default.
- W1770758908 hasRelatedWork W2384129116 @default.
- W1770758908 hasRelatedWork W2766721049 @default.
- W1770758908 hasRelatedWork W2961085424 @default.
- W1770758908 hasRelatedWork W3145924829 @default.
- W1770758908 hasRelatedWork W3152267458 @default.
- W1770758908 isParatext "false" @default.
- W1770758908 isRetracted "false" @default.
- W1770758908 magId "1770758908" @default.
- W1770758908 workType "article" @default.