Matches in SemOpenAlex for { <https://semopenalex.org/work/W4372262650> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W4372262650 abstract "Transformer-based end-to-end speech recognition has achieved great success. However, the large footprint and computational overhead make it difficult to deploy these models in some real-world applications. Model compression techniques can reduce the model size and speed up inference, but the compressed model has a fixed architecture which might be suboptimal. We propose a novel Transformer encoder with Input-Dependent Dynamic Depth (I3D) to achieve strong performance-efficiency trade-offs. With a similar number of layers at inference time, I3D-based models outperform the vanilla Transformer and the static pruned model via iterative layer pruning. We also present interesting analysis on the gate probabilities and the input-dependency, which helps us better understand deep encoders." @default.
- W4372262650 created "2023-05-07" @default.
- W4372262650 creator A5001291873 @default.
- W4372262650 creator A5015578080 @default.
- W4372262650 creator A5087513674 @default.
- W4372262650 date "2023-06-04" @default.
- W4372262650 modified "2023-09-27" @default.
- W4372262650 title "I3D: Transformer Architectures with Input-Dependent Dynamic Depth for Speech Recognition" @default.
- W4372262650 cites W1494198834 @default.
- W4372262650 cites W2127141656 @default.
- W4372262650 cites W2157331557 @default.
- W4372262650 cites W2327501763 @default.
- W4372262650 cites W2884751099 @default.
- W4372262650 cites W2962780374 @default.
- W4372262650 cites W2962944050 @default.
- W4372262650 cites W2963393494 @default.
- W4372262650 cites W2972818416 @default.
- W4372262650 cites W3037913581 @default.
- W4372262650 cites W3091900426 @default.
- W4372262650 cites W3097777922 @default.
- W4372262650 cites W3151287998 @default.
- W4372262650 cites W3162249256 @default.
- W4372262650 cites W3163105696 @default.
- W4372262650 cites W3173563887 @default.
- W4372262650 cites W3196783077 @default.
- W4372262650 cites W3197148831 @default.
- W4372262650 cites W3198368663 @default.
- W4372262650 cites W3200669029 @default.
- W4372262650 cites W3203140070 @default.
- W4372262650 cites W3204647170 @default.
- W4372262650 cites W4224821750 @default.
- W4372262650 cites W4225302959 @default.
- W4372262650 cites W4284881636 @default.
- W4372262650 doi "https://doi.org/10.1109/icassp49357.2023.10096662" @default.
- W4372262650 hasPublicationYear "2023" @default.
- W4372262650 type Work @default.
- W4372262650 citedByCount "0" @default.
- W4372262650 crossrefType "proceedings-article" @default.
- W4372262650 hasAuthorship W4372262650A5001291873 @default.
- W4372262650 hasAuthorship W4372262650A5015578080 @default.
- W4372262650 hasAuthorship W4372262650A5087513674 @default.
- W4372262650 hasBestOaLocation W43722626501 @default.
- W4372262650 hasConcept C111919701 @default.
- W4372262650 hasConcept C118505674 @default.
- W4372262650 hasConcept C121332964 @default.
- W4372262650 hasConcept C154945302 @default.
- W4372262650 hasConcept C165801399 @default.
- W4372262650 hasConcept C2776214188 @default.
- W4372262650 hasConcept C28490314 @default.
- W4372262650 hasConcept C41008148 @default.
- W4372262650 hasConcept C62520636 @default.
- W4372262650 hasConcept C66322947 @default.
- W4372262650 hasConcept C74912251 @default.
- W4372262650 hasConceptScore W4372262650C111919701 @default.
- W4372262650 hasConceptScore W4372262650C118505674 @default.
- W4372262650 hasConceptScore W4372262650C121332964 @default.
- W4372262650 hasConceptScore W4372262650C154945302 @default.
- W4372262650 hasConceptScore W4372262650C165801399 @default.
- W4372262650 hasConceptScore W4372262650C2776214188 @default.
- W4372262650 hasConceptScore W4372262650C28490314 @default.
- W4372262650 hasConceptScore W4372262650C41008148 @default.
- W4372262650 hasConceptScore W4372262650C62520636 @default.
- W4372262650 hasConceptScore W4372262650C66322947 @default.
- W4372262650 hasConceptScore W4372262650C74912251 @default.
- W4372262650 hasFunder F4320306076 @default.
- W4372262650 hasFunder F4320307110 @default.
- W4372262650 hasLocation W43722626501 @default.
- W4372262650 hasLocation W43722626502 @default.
- W4372262650 hasOpenAccess W4372262650 @default.
- W4372262650 hasPrimaryLocation W43722626501 @default.
- W4372262650 hasRelatedWork W2275988210 @default.
- W4372262650 hasRelatedWork W2368779261 @default.
- W4372262650 hasRelatedWork W2385621972 @default.
- W4372262650 hasRelatedWork W2589098947 @default.
- W4372262650 hasRelatedWork W2892009249 @default.
- W4372262650 hasRelatedWork W4206178588 @default.
- W4372262650 hasRelatedWork W4287635093 @default.
- W4372262650 hasRelatedWork W4307334198 @default.
- W4372262650 hasRelatedWork W4319995060 @default.
- W4372262650 hasRelatedWork W4385825655 @default.
- W4372262650 isParatext "false" @default.
- W4372262650 isRetracted "false" @default.
- W4372262650 workType "article" @default.