Matches in SemOpenAlex for { <https://semopenalex.org/work/W2953242772> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W2953242772 abstract "Recurrent Neural Networks (RNNs) have dominated language modeling because of their superior performance over traditional N-gram based models. In many applications, a large Recurrent Neural Network language model (RNNLM) or an ensemble of several RNNLMs is used. These models have large memory footprints and require heavy computation. In this paper, we examine the effect of applying knowledge distillation in reducing the model size for RNNLMs. In addition, we propose a trust regularization method to improve the knowledge distillation training for RNNLMs. Using knowledge distillation with trust regularization, we reduce the parameter size to a third of that of the previously published best model while maintaining the state-of-the-art perplexity result on Penn Treebank data. In a speech recognition N-bestrescoring task, we reduce the RNNLM model size to 18.5% of the baseline system, with no degradation in word error rate(WER) performance on Wall Street Journal data set." @default.
- W2953242772 created "2019-06-27" @default.
- W2953242772 creator A5008349950 @default.
- W2953242772 creator A5030660132 @default.
- W2953242772 creator A5049457213 @default.
- W2953242772 creator A5085953379 @default.
- W2953242772 date "2019-04-08" @default.
- W2953242772 modified "2023-10-03" @default.
- W2953242772 title "Knowledge Distillation For Recurrent Neural Network Language Modeling With Trust Regularization" @default.
- W2953242772 cites W1524333225 @default.
- W2953242772 cites W1591801644 @default.
- W2953242772 cites W1690739335 @default.
- W2953242772 cites W179875071 @default.
- W2953242772 cites W1821462560 @default.
- W2953242772 cites W1938755728 @default.
- W2953242772 cites W2024490156 @default.
- W2953242772 cites W2250379827 @default.
- W2953242772 cites W2286365479 @default.
- W2953242772 cites W2294543795 @default.
- W2953242772 cites W2951672049 @default.
- W2953242772 cites W2952899695 @default.
- W2953242772 cites W2962832505 @default.
- W2953242772 cites W2962964385 @default.
- W2953242772 cites W2962965405 @default.
- W2953242772 cites W2963266340 @default.
- W2953242772 cites W2963374479 @default.
- W2953242772 cites W2963494889 @default.
- W2953242772 cites W2963537482 @default.
- W2953242772 cites W2963643655 @default.
- W2953242772 cites W2963748792 @default.
- W2953242772 cites W2963983719 @default.
- W2953242772 cites W4919037 @default.
- W2953242772 doi "https://doi.org/10.48550/arxiv.1904.04163" @default.
- W2953242772 hasPublicationYear "2019" @default.
- W2953242772 type Work @default.
- W2953242772 sameAs 2953242772 @default.
- W2953242772 citedByCount "0" @default.
- W2953242772 crossrefType "posted-content" @default.
- W2953242772 hasAuthorship W2953242772A5008349950 @default.
- W2953242772 hasAuthorship W2953242772A5030660132 @default.
- W2953242772 hasAuthorship W2953242772A5049457213 @default.
- W2953242772 hasAuthorship W2953242772A5085953379 @default.
- W2953242772 hasBestOaLocation W29532427721 @default.
- W2953242772 hasConcept C100279451 @default.
- W2953242772 hasConcept C119857082 @default.
- W2953242772 hasConcept C137293760 @default.
- W2953242772 hasConcept C147168706 @default.
- W2953242772 hasConcept C154945302 @default.
- W2953242772 hasConcept C169903167 @default.
- W2953242772 hasConcept C178790620 @default.
- W2953242772 hasConcept C185592680 @default.
- W2953242772 hasConcept C19768560 @default.
- W2953242772 hasConcept C204030448 @default.
- W2953242772 hasConcept C204321447 @default.
- W2953242772 hasConcept C206134035 @default.
- W2953242772 hasConcept C2776135515 @default.
- W2953242772 hasConcept C40969351 @default.
- W2953242772 hasConcept C41008148 @default.
- W2953242772 hasConcept C50644808 @default.
- W2953242772 hasConceptScore W2953242772C100279451 @default.
- W2953242772 hasConceptScore W2953242772C119857082 @default.
- W2953242772 hasConceptScore W2953242772C137293760 @default.
- W2953242772 hasConceptScore W2953242772C147168706 @default.
- W2953242772 hasConceptScore W2953242772C154945302 @default.
- W2953242772 hasConceptScore W2953242772C169903167 @default.
- W2953242772 hasConceptScore W2953242772C178790620 @default.
- W2953242772 hasConceptScore W2953242772C185592680 @default.
- W2953242772 hasConceptScore W2953242772C19768560 @default.
- W2953242772 hasConceptScore W2953242772C204030448 @default.
- W2953242772 hasConceptScore W2953242772C204321447 @default.
- W2953242772 hasConceptScore W2953242772C206134035 @default.
- W2953242772 hasConceptScore W2953242772C2776135515 @default.
- W2953242772 hasConceptScore W2953242772C40969351 @default.
- W2953242772 hasConceptScore W2953242772C41008148 @default.
- W2953242772 hasConceptScore W2953242772C50644808 @default.
- W2953242772 hasLocation W29532427721 @default.
- W2953242772 hasOpenAccess W2953242772 @default.
- W2953242772 hasPrimaryLocation W29532427721 @default.
- W2953242772 hasRelatedWork W2072350286 @default.
- W2953242772 hasRelatedWork W2167980204 @default.
- W2953242772 hasRelatedWork W2251150025 @default.
- W2953242772 hasRelatedWork W2508661145 @default.
- W2953242772 hasRelatedWork W2598933668 @default.
- W2953242772 hasRelatedWork W2638027694 @default.
- W2953242772 hasRelatedWork W2743945814 @default.
- W2953242772 hasRelatedWork W2953242772 @default.
- W2953242772 hasRelatedWork W2962832505 @default.
- W2953242772 hasRelatedWork W4299838440 @default.
- W2953242772 isParatext "false" @default.
- W2953242772 isRetracted "false" @default.
- W2953242772 magId "2953242772" @default.
- W2953242772 workType "article" @default.