Matches in SemOpenAlex for { <https://semopenalex.org/work/W2471933213> ?p ?o ?g. }
- W2471933213 abstract "We present a comprehensive study of deep bidirectional long short-term memory (LSTM) recurrent neural network (RNN) based acoustic models for automatic speech recognition (ASR). We study the effect of size and depth and train models of up to 8 layers. We investigate the training aspect and study different variants of optimization methods, batching, truncated backpropagation, different regularization techniques such as dropout and $L_2$ regularization, and different gradient clipping variants. The major part of the experimental analysis was performed on the Quaero corpus. Additional experiments also were performed on the Switchboard corpus. Our best LSTM model has a relative improvement in word error rate of over 14% compared to our best feed-forward neural network (FFNN) baseline on the Quaero task. On this task, we get our best result with an 8 layer bidirectional LSTM and we show that a pretraining scheme with layer-wise construction helps for deep LSTMs. Finally we compare the training calculation time of many of the presented experiments in relation with recognition performance. All the experiments were done with RETURNN, the RWTH extensible training framework for universal recurrent neural networks in combination with RASR, the RWTH ASR toolkit." @default.
- W2471933213 created "2016-07-22" @default.
- W2471933213 creator A5041202059 @default.
- W2471933213 creator A5059521407 @default.
- W2471933213 creator A5082582385 @default.
- W2471933213 creator A5087367411 @default.
- W2471933213 creator A5088968292 @default.
- W2471933213 date "2017-03-01" @default.
- W2471933213 modified "2023-10-17" @default.
- W2471933213 title "A comprehensive study of deep bidirectional LSTM RNNS for acoustic modeling in speech recognition" @default.
- W2471933213 cites W1525636403 @default.
- W2471933213 cites W1600744878 @default.
- W2471933213 cites W1886418050 @default.
- W2471933213 cites W1988720110 @default.
- W2471933213 cites W2005708641 @default.
- W2471933213 cites W2024200390 @default.
- W2471933213 cites W2064675550 @default.
- W2471933213 cites W2076063813 @default.
- W2471933213 cites W2079735306 @default.
- W2471933213 cites W2084894614 @default.
- W2471933213 cites W2110871230 @default.
- W2471933213 cites W2119615570 @default.
- W2471933213 cites W2131774270 @default.
- W2471933213 cites W2160306971 @default.
- W2471933213 cites W2291282366 @default.
- W2471933213 cites W2512608784 @default.
- W2471933213 cites W811578723 @default.
- W2471933213 doi "https://doi.org/10.1109/icassp.2017.7952599" @default.
- W2471933213 hasPublicationYear "2017" @default.
- W2471933213 type Work @default.
- W2471933213 sameAs 2471933213 @default.
- W2471933213 citedByCount "110" @default.
- W2471933213 countsByYear W24719332132016 @default.
- W2471933213 countsByYear W24719332132017 @default.
- W2471933213 countsByYear W24719332132018 @default.
- W2471933213 countsByYear W24719332132019 @default.
- W2471933213 countsByYear W24719332132020 @default.
- W2471933213 countsByYear W24719332132021 @default.
- W2471933213 countsByYear W24719332132022 @default.
- W2471933213 countsByYear W24719332132023 @default.
- W2471933213 crossrefType "proceedings-article" @default.
- W2471933213 hasAuthorship W2471933213A5041202059 @default.
- W2471933213 hasAuthorship W2471933213A5059521407 @default.
- W2471933213 hasAuthorship W2471933213A5082582385 @default.
- W2471933213 hasAuthorship W2471933213A5087367411 @default.
- W2471933213 hasAuthorship W2471933213A5088968292 @default.
- W2471933213 hasBestOaLocation W24719332132 @default.
- W2471933213 hasConcept C108583219 @default.
- W2471933213 hasConcept C119857082 @default.
- W2471933213 hasConcept C127413603 @default.
- W2471933213 hasConcept C133731056 @default.
- W2471933213 hasConcept C137293760 @default.
- W2471933213 hasConcept C147168706 @default.
- W2471933213 hasConcept C154945302 @default.
- W2471933213 hasConcept C155032097 @default.
- W2471933213 hasConcept C155635449 @default.
- W2471933213 hasConcept C162324750 @default.
- W2471933213 hasConcept C187736073 @default.
- W2471933213 hasConcept C2776135515 @default.
- W2471933213 hasConcept C2776145597 @default.
- W2471933213 hasConcept C2780451532 @default.
- W2471933213 hasConcept C28490314 @default.
- W2471933213 hasConcept C38858127 @default.
- W2471933213 hasConcept C40969351 @default.
- W2471933213 hasConcept C41008148 @default.
- W2471933213 hasConcept C50644808 @default.
- W2471933213 hasConcept C61328038 @default.
- W2471933213 hasConceptScore W2471933213C108583219 @default.
- W2471933213 hasConceptScore W2471933213C119857082 @default.
- W2471933213 hasConceptScore W2471933213C127413603 @default.
- W2471933213 hasConceptScore W2471933213C133731056 @default.
- W2471933213 hasConceptScore W2471933213C137293760 @default.
- W2471933213 hasConceptScore W2471933213C147168706 @default.
- W2471933213 hasConceptScore W2471933213C154945302 @default.
- W2471933213 hasConceptScore W2471933213C155032097 @default.
- W2471933213 hasConceptScore W2471933213C155635449 @default.
- W2471933213 hasConceptScore W2471933213C162324750 @default.
- W2471933213 hasConceptScore W2471933213C187736073 @default.
- W2471933213 hasConceptScore W2471933213C2776135515 @default.
- W2471933213 hasConceptScore W2471933213C2776145597 @default.
- W2471933213 hasConceptScore W2471933213C2780451532 @default.
- W2471933213 hasConceptScore W2471933213C28490314 @default.
- W2471933213 hasConceptScore W2471933213C38858127 @default.
- W2471933213 hasConceptScore W2471933213C40969351 @default.
- W2471933213 hasConceptScore W2471933213C41008148 @default.
- W2471933213 hasConceptScore W2471933213C50644808 @default.
- W2471933213 hasConceptScore W2471933213C61328038 @default.
- W2471933213 hasLocation W24719332131 @default.
- W2471933213 hasLocation W24719332132 @default.
- W2471933213 hasOpenAccess W2471933213 @default.
- W2471933213 hasPrimaryLocation W24719332131 @default.
- W2471933213 hasRelatedWork W1872348600 @default.
- W2471933213 hasRelatedWork W2244609359 @default.
- W2471933213 hasRelatedWork W2471933213 @default.
- W2471933213 hasRelatedWork W2512056531 @default.
- W2471933213 hasRelatedWork W2782005958 @default.
- W2471933213 hasRelatedWork W2786709961 @default.
- W2471933213 hasRelatedWork W2843351054 @default.
- W2471933213 hasRelatedWork W2949937978 @default.
- W2471933213 hasRelatedWork W4205868073 @default.