Matches in SemOpenAlex for { <https://semopenalex.org/work/W1984985727> ?p ?o ?g. }
- W1984985727 abstract "We address the problem of subselecting a large set of acoustic data to train automatic speech recognition (ASR) systems. To this end, we apply a novel data selection technique based on constrained submodular function maximization. Though NP-hard, the combinatorial optimization problem can be approximately solved by a simple and scalable greedy algorithm with constant-factor guarantees. We evaluate our approach by subselecting data from 1300 hours of conversational English telephone data to train two types large-vocabulary speech recognizers, one with Gaussian mixture model (GMM) based acoustic models, and another based on deep neural networks (DNNs). We show that training data can be reduced significantly, and that our technique outperforms both random selection and a previously proposed selection method utilizing comparable resources. Notably, using the submodular selection method, the DNN system using only about 5% of the training data is able to achieve performance on par with the GMM system using 100% of the training data - with the baseline subset selection methods, however, the DNN system is unable to accomplish this correspondence." @default.
- W1984985727 created "2016-06-24" @default.
- W1984985727 creator A5000919546 @default.
- W1984985727 creator A5031137004 @default.
- W1984985727 creator A5043149936 @default.
- W1984985727 creator A5050325468 @default.
- W1984985727 creator A5091674429 @default.
- W1984985727 date "2014-05-01" @default.
- W1984985727 modified "2023-10-01" @default.
- W1984985727 title "Submodular subset selection for large-scale speech training data" @default.
- W1984985727 cites W1680189815 @default.
- W1984985727 cites W1898824936 @default.
- W1984985727 cites W195739637 @default.
- W1984985727 cites W1975113979 @default.
- W1984985727 cites W2032020816 @default.
- W1984985727 cites W2045563462 @default.
- W1984985727 cites W2053625667 @default.
- W1984985727 cites W2125610823 @default.
- W1984985727 cites W2135241496 @default.
- W1984985727 cites W2143996311 @default.
- W1984985727 cites W2161482971 @default.
- W1984985727 cites W3149335959 @default.
- W1984985727 doi "https://doi.org/10.1109/icassp.2014.6854213" @default.
- W1984985727 hasPublicationYear "2014" @default.
- W1984985727 type Work @default.
- W1984985727 sameAs 1984985727 @default.
- W1984985727 citedByCount "55" @default.
- W1984985727 countsByYear W19849857272014 @default.
- W1984985727 countsByYear W19849857272015 @default.
- W1984985727 countsByYear W19849857272016 @default.
- W1984985727 countsByYear W19849857272017 @default.
- W1984985727 countsByYear W19849857272018 @default.
- W1984985727 countsByYear W19849857272019 @default.
- W1984985727 countsByYear W19849857272020 @default.
- W1984985727 countsByYear W19849857272021 @default.
- W1984985727 countsByYear W19849857272022 @default.
- W1984985727 crossrefType "proceedings-article" @default.
- W1984985727 hasAuthorship W1984985727A5000919546 @default.
- W1984985727 hasAuthorship W1984985727A5031137004 @default.
- W1984985727 hasAuthorship W1984985727A5043149936 @default.
- W1984985727 hasAuthorship W1984985727A5050325468 @default.
- W1984985727 hasAuthorship W1984985727A5091674429 @default.
- W1984985727 hasConcept C11413529 @default.
- W1984985727 hasConcept C119857082 @default.
- W1984985727 hasConcept C121332964 @default.
- W1984985727 hasConcept C126255220 @default.
- W1984985727 hasConcept C138885662 @default.
- W1984985727 hasConcept C148483581 @default.
- W1984985727 hasConcept C154945302 @default.
- W1984985727 hasConcept C163716315 @default.
- W1984985727 hasConcept C178621042 @default.
- W1984985727 hasConcept C23224414 @default.
- W1984985727 hasConcept C2776330181 @default.
- W1984985727 hasConcept C2777601683 @default.
- W1984985727 hasConcept C28490314 @default.
- W1984985727 hasConcept C33923547 @default.
- W1984985727 hasConcept C41008148 @default.
- W1984985727 hasConcept C41895202 @default.
- W1984985727 hasConcept C48044578 @default.
- W1984985727 hasConcept C51823790 @default.
- W1984985727 hasConcept C61224824 @default.
- W1984985727 hasConcept C61326573 @default.
- W1984985727 hasConcept C62520636 @default.
- W1984985727 hasConcept C77088390 @default.
- W1984985727 hasConcept C81917197 @default.
- W1984985727 hasConceptScore W1984985727C11413529 @default.
- W1984985727 hasConceptScore W1984985727C119857082 @default.
- W1984985727 hasConceptScore W1984985727C121332964 @default.
- W1984985727 hasConceptScore W1984985727C126255220 @default.
- W1984985727 hasConceptScore W1984985727C138885662 @default.
- W1984985727 hasConceptScore W1984985727C148483581 @default.
- W1984985727 hasConceptScore W1984985727C154945302 @default.
- W1984985727 hasConceptScore W1984985727C163716315 @default.
- W1984985727 hasConceptScore W1984985727C178621042 @default.
- W1984985727 hasConceptScore W1984985727C23224414 @default.
- W1984985727 hasConceptScore W1984985727C2776330181 @default.
- W1984985727 hasConceptScore W1984985727C2777601683 @default.
- W1984985727 hasConceptScore W1984985727C28490314 @default.
- W1984985727 hasConceptScore W1984985727C33923547 @default.
- W1984985727 hasConceptScore W1984985727C41008148 @default.
- W1984985727 hasConceptScore W1984985727C41895202 @default.
- W1984985727 hasConceptScore W1984985727C48044578 @default.
- W1984985727 hasConceptScore W1984985727C51823790 @default.
- W1984985727 hasConceptScore W1984985727C61224824 @default.
- W1984985727 hasConceptScore W1984985727C61326573 @default.
- W1984985727 hasConceptScore W1984985727C62520636 @default.
- W1984985727 hasConceptScore W1984985727C77088390 @default.
- W1984985727 hasConceptScore W1984985727C81917197 @default.
- W1984985727 hasLocation W19849857271 @default.
- W1984985727 hasOpenAccess W1984985727 @default.
- W1984985727 hasPrimaryLocation W19849857271 @default.
- W1984985727 hasRelatedWork W2112634534 @default.
- W1984985727 hasRelatedWork W2144629587 @default.
- W1984985727 hasRelatedWork W2159208002 @default.
- W1984985727 hasRelatedWork W2168237838 @default.
- W1984985727 hasRelatedWork W2413963148 @default.
- W1984985727 hasRelatedWork W2545766074 @default.
- W1984985727 hasRelatedWork W2949672782 @default.
- W1984985727 hasRelatedWork W4294276810 @default.
- W1984985727 hasRelatedWork W4298470087 @default.