Matches in SemOpenAlex for { <https://semopenalex.org/work/W4319862233> ?p ?o ?g. }
Showing items 1 to 90 of
90
with 100 items per page.
- W4319862233 abstract "End-to-end automatic lip-reading usually comprises an encoder-decoder model and an optional external language model. In this work, we introduce two regularization methods to the field of lip-reading: First, we apply the regularized dropout (R-Drop) method to transformer-based lip-reading to improve their training-inference consistency. Second, the relaxed attention technique is applied during training for a better external language model integration. We are the first to show that these two complementary approaches yield particu1arly strong performance if combined in the right manner. In particular, by adding an additional R - Drop loss and smoothing the attention weights in cross multi-head attention during training only, we achieve a new state of the art with a word error rate of 22.2% on Lip Reading Sentences 2 (LRS2). On LRS3, we are 2nd ranked with 25.5% WER using only 1,759 h of training data, while the 1 st rank uses about 90,000 h. Our code is available at GitHub. <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>1</sup> <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>1</sup> https://github.com/ifnspaml/Lipreading-RDrop-RA" @default.
- W4319862233 created "2023-02-11" @default.
- W4319862233 creator A5002593702 @default.
- W4319862233 creator A5039030019 @default.
- W4319862233 creator A5052937699 @default.
- W4319862233 creator A5067100305 @default.
- W4319862233 date "2023-01-09" @default.
- W4319862233 modified "2023-10-18" @default.
- W4319862233 title "Transformer-Based Lip-Reading with Regularized Dropout and Relaxed Attention" @default.
- W4319862233 cites W1494198834 @default.
- W4319862233 cites W2526425061 @default.
- W4319862233 cites W2577366047 @default.
- W4319862233 cites W2808631503 @default.
- W4319862233 cites W2888779557 @default.
- W4319862233 cites W2892009249 @default.
- W4319862233 cites W2912984882 @default.
- W4319862233 cites W2952746495 @default.
- W4319862233 cites W2963250244 @default.
- W4319862233 cites W2963362078 @default.
- W4319862233 cites W2963528589 @default.
- W4319862233 cites W2963654155 @default.
- W4319862233 cites W3006974783 @default.
- W4319862233 cites W3008037978 @default.
- W4319862233 cites W3015830103 @default.
- W4319862233 cites W3016011581 @default.
- W4319862233 cites W3096318498 @default.
- W4319862233 cites W3097777922 @default.
- W4319862233 cites W3162293946 @default.
- W4319862233 cites W3163169798 @default.
- W4319862233 cites W3197567540 @default.
- W4319862233 cites W3198871389 @default.
- W4319862233 cites W4307286264 @default.
- W4319862233 cites W4312638101 @default.
- W4319862233 doi "https://doi.org/10.1109/slt54892.2023.10023442" @default.
- W4319862233 hasPublicationYear "2023" @default.
- W4319862233 type Work @default.
- W4319862233 citedByCount "0" @default.
- W4319862233 crossrefType "proceedings-article" @default.
- W4319862233 hasAuthorship W4319862233A5002593702 @default.
- W4319862233 hasAuthorship W4319862233A5039030019 @default.
- W4319862233 hasAuthorship W4319862233A5052937699 @default.
- W4319862233 hasAuthorship W4319862233A5067100305 @default.
- W4319862233 hasConcept C111919701 @default.
- W4319862233 hasConcept C11413529 @default.
- W4319862233 hasConcept C118505674 @default.
- W4319862233 hasConcept C121332964 @default.
- W4319862233 hasConcept C154945302 @default.
- W4319862233 hasConcept C165801399 @default.
- W4319862233 hasConcept C204321447 @default.
- W4319862233 hasConcept C2776135515 @default.
- W4319862233 hasConcept C2776214188 @default.
- W4319862233 hasConcept C28490314 @default.
- W4319862233 hasConcept C31972630 @default.
- W4319862233 hasConcept C3770464 @default.
- W4319862233 hasConcept C41008148 @default.
- W4319862233 hasConcept C57273362 @default.
- W4319862233 hasConcept C62520636 @default.
- W4319862233 hasConcept C66322947 @default.
- W4319862233 hasConceptScore W4319862233C111919701 @default.
- W4319862233 hasConceptScore W4319862233C11413529 @default.
- W4319862233 hasConceptScore W4319862233C118505674 @default.
- W4319862233 hasConceptScore W4319862233C121332964 @default.
- W4319862233 hasConceptScore W4319862233C154945302 @default.
- W4319862233 hasConceptScore W4319862233C165801399 @default.
- W4319862233 hasConceptScore W4319862233C204321447 @default.
- W4319862233 hasConceptScore W4319862233C2776135515 @default.
- W4319862233 hasConceptScore W4319862233C2776214188 @default.
- W4319862233 hasConceptScore W4319862233C28490314 @default.
- W4319862233 hasConceptScore W4319862233C31972630 @default.
- W4319862233 hasConceptScore W4319862233C3770464 @default.
- W4319862233 hasConceptScore W4319862233C41008148 @default.
- W4319862233 hasConceptScore W4319862233C57273362 @default.
- W4319862233 hasConceptScore W4319862233C62520636 @default.
- W4319862233 hasConceptScore W4319862233C66322947 @default.
- W4319862233 hasLocation W43198622331 @default.
- W4319862233 hasOpenAccess W4319862233 @default.
- W4319862233 hasPrimaryLocation W43198622331 @default.
- W4319862233 hasRelatedWork W1950712214 @default.
- W4319862233 hasRelatedWork W2892009249 @default.
- W4319862233 hasRelatedWork W2898132662 @default.
- W4319862233 hasRelatedWork W2992696780 @default.
- W4319862233 hasRelatedWork W3116268265 @default.
- W4319862233 hasRelatedWork W3156915121 @default.
- W4319862233 hasRelatedWork W3197792581 @default.
- W4319862233 hasRelatedWork W3207932232 @default.
- W4319862233 hasRelatedWork W4286982949 @default.
- W4319862233 hasRelatedWork W4316116709 @default.
- W4319862233 isParatext "false" @default.
- W4319862233 isRetracted "false" @default.
- W4319862233 workType "article" @default.