Matches in SemOpenAlex for { <https://semopenalex.org/work/W2034261344> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W2034261344 abstract "We describe a hybrid GPU/CPU architecture for stochastic gradient descent training of neural network acoustic models under a lattice-based minimum Bayes risk (MBR) criterion. The crux of the method is to run SGD on a GPU card which consumes framerandomized mini-batches produced by multiple workers running on a cluster of multi-core CPU nodes which compute HMM state MBR occupancies. To minimize communication cost, a separate thread running on the GPU host receives minibatches from and sends updated models to the workers, and communicates with the SGD thread via a producer-consumer queue of minibatches. Using this architecture, it is possible to match the speed of GPU-based SGD cross-entropy (CE) training (1 hour of processing per 100 hours of audio on Switchboard). Additionally, we compare different ways of doing frame randomization and discuss experimental results on three LVCSR tasks (Switchboard 300 hours, English broadcast news 50 hours, and noisy Levantine telephone conversations 300 hours)." @default.
- W2034261344 created "2016-06-24" @default.
- W2034261344 creator A5079994647 @default.
- W2034261344 date "2014-12-01" @default.
- W2034261344 modified "2023-09-25" @default.
- W2034261344 title "A distributed architecture for fast SGD sequence discriminative training of DNN acoustic models" @default.
- W2034261344 cites W1985371235 @default.
- W2034261344 cites W1987238397 @default.
- W2034261344 cites W2028706510 @default.
- W2034261344 cites W2035846950 @default.
- W2034261344 cites W2058641082 @default.
- W2034261344 cites W2079623482 @default.
- W2034261344 cites W2087402357 @default.
- W2034261344 cites W2094233035 @default.
- W2034261344 cites W2114016253 @default.
- W2034261344 cites W2131342762 @default.
- W2034261344 cites W2160306971 @default.
- W2034261344 cites W2160815625 @default.
- W2034261344 cites W2168231600 @default.
- W2034261344 cites W2294723541 @default.
- W2034261344 cites W2403195671 @default.
- W2034261344 cites W2405787239 @default.
- W2034261344 doi "https://doi.org/10.1109/slt.2014.7078571" @default.
- W2034261344 hasPublicationYear "2014" @default.
- W2034261344 type Work @default.
- W2034261344 sameAs 2034261344 @default.
- W2034261344 citedByCount "0" @default.
- W2034261344 crossrefType "proceedings-article" @default.
- W2034261344 hasAuthorship W2034261344A5079994647 @default.
- W2034261344 hasConcept C123657996 @default.
- W2034261344 hasConcept C142362112 @default.
- W2034261344 hasConcept C153349607 @default.
- W2034261344 hasConcept C154945302 @default.
- W2034261344 hasConcept C160403385 @default.
- W2034261344 hasConcept C206688291 @default.
- W2034261344 hasConcept C23224414 @default.
- W2034261344 hasConcept C28490314 @default.
- W2034261344 hasConcept C31258907 @default.
- W2034261344 hasConcept C41008148 @default.
- W2034261344 hasConcept C50644808 @default.
- W2034261344 hasConcept C97931131 @default.
- W2034261344 hasConceptScore W2034261344C123657996 @default.
- W2034261344 hasConceptScore W2034261344C142362112 @default.
- W2034261344 hasConceptScore W2034261344C153349607 @default.
- W2034261344 hasConceptScore W2034261344C154945302 @default.
- W2034261344 hasConceptScore W2034261344C160403385 @default.
- W2034261344 hasConceptScore W2034261344C206688291 @default.
- W2034261344 hasConceptScore W2034261344C23224414 @default.
- W2034261344 hasConceptScore W2034261344C28490314 @default.
- W2034261344 hasConceptScore W2034261344C31258907 @default.
- W2034261344 hasConceptScore W2034261344C41008148 @default.
- W2034261344 hasConceptScore W2034261344C50644808 @default.
- W2034261344 hasConceptScore W2034261344C97931131 @default.
- W2034261344 hasLocation W20342613441 @default.
- W2034261344 hasOpenAccess W2034261344 @default.
- W2034261344 hasPrimaryLocation W20342613441 @default.
- W2034261344 hasRelatedWork W1639373162 @default.
- W2034261344 hasRelatedWork W1965154800 @default.
- W2034261344 hasRelatedWork W1998656041 @default.
- W2034261344 hasRelatedWork W2035846950 @default.
- W2034261344 hasRelatedWork W2064750458 @default.
- W2034261344 hasRelatedWork W2071515987 @default.
- W2034261344 hasRelatedWork W2094233035 @default.
- W2034261344 hasRelatedWork W2292744640 @default.
- W2034261344 hasRelatedWork W2294723541 @default.
- W2034261344 hasRelatedWork W2346091587 @default.
- W2034261344 hasRelatedWork W2353573803 @default.
- W2034261344 hasRelatedWork W2405883473 @default.
- W2034261344 hasRelatedWork W2488202690 @default.
- W2034261344 hasRelatedWork W2936012302 @default.
- W2034261344 hasRelatedWork W2964265128 @default.
- W2034261344 hasRelatedWork W2971871332 @default.
- W2034261344 hasRelatedWork W3016159127 @default.
- W2034261344 hasRelatedWork W3105037130 @default.
- W2034261344 hasRelatedWork W1990471071 @default.
- W2034261344 hasRelatedWork W3207935346 @default.
- W2034261344 isParatext "false" @default.
- W2034261344 isRetracted "false" @default.
- W2034261344 magId "2034261344" @default.
- W2034261344 workType "article" @default.