Matches in SemOpenAlex for { <https://semopenalex.org/work/W4224919704> ?p ?o ?g. }
Showing items 1 to 92 of
92
with 100 items per page.
- W4224919704 abstract "This paper introduces contrastive siamese (c-siam) network, an architecture for leveraging unlabeled acoustic data in speech recognition. c-siam is the first network that extracts high-level linguistic information from speech by matching outputs of two identical transformer encoders. It contains augmented and target branches which are trained by: (1) masking inputs and matching outputs with a contrastive loss, (2) incorporating a stop gradient operation on the target branch, (3) using an extra learnable transformation on the augmented branch, (4) introducing new temporal augment functions to prevent the shortcut learning problem. We use the Libri-light 60k unsupervised data and the LibriSpeech 100hrs/960hrs supervised data to compare c-siam and other best-performing systems. Our experiments show that c-siam provides 20% relative word error rate improvement over wav2vec baselines. A c-siam network with 450M parameters achieves competitive results compared to the state-of-the-art networks with 600M parameters." @default.
- W4224919704 created "2022-04-28" @default.
- W4224919704 creator A5001053064 @default.
- W4224919704 creator A5017569678 @default.
- W4224919704 creator A5025206583 @default.
- W4224919704 creator A5052441498 @default.
- W4224919704 creator A5072978012 @default.
- W4224919704 creator A5086894713 @default.
- W4224919704 date "2022-05-23" @default.
- W4224919704 modified "2023-10-18" @default.
- W4224919704 title "Contrastive Siamese Network for Semi-Supervised Speech Recognition" @default.
- W4224919704 cites W1494198834 @default.
- W4224919704 cites W2168510624 @default.
- W4224919704 cites W2936774411 @default.
- W4224919704 cites W2964110616 @default.
- W4224919704 cites W2972943112 @default.
- W4224919704 cites W2973049979 @default.
- W4224919704 cites W2982223350 @default.
- W4224919704 cites W3015265920 @default.
- W4224919704 cites W3016010032 @default.
- W4224919704 cites W3016970897 @default.
- W4224919704 cites W3035524453 @default.
- W4224919704 cites W3041561163 @default.
- W4224919704 cites W3097777922 @default.
- W4224919704 cites W3171007011 @default.
- W4224919704 cites W4226033575 @default.
- W4224919704 doi "https://doi.org/10.1109/icassp43922.2022.9747355" @default.
- W4224919704 hasPublicationYear "2022" @default.
- W4224919704 type Work @default.
- W4224919704 citedByCount "3" @default.
- W4224919704 countsByYear W42249197042023 @default.
- W4224919704 crossrefType "proceedings-article" @default.
- W4224919704 hasAuthorship W4224919704A5001053064 @default.
- W4224919704 hasAuthorship W4224919704A5017569678 @default.
- W4224919704 hasAuthorship W4224919704A5025206583 @default.
- W4224919704 hasAuthorship W4224919704A5052441498 @default.
- W4224919704 hasAuthorship W4224919704A5072978012 @default.
- W4224919704 hasAuthorship W4224919704A5086894713 @default.
- W4224919704 hasConcept C105795698 @default.
- W4224919704 hasConcept C111919701 @default.
- W4224919704 hasConcept C118505674 @default.
- W4224919704 hasConcept C121332964 @default.
- W4224919704 hasConcept C142362112 @default.
- W4224919704 hasConcept C153180895 @default.
- W4224919704 hasConcept C153349607 @default.
- W4224919704 hasConcept C154945302 @default.
- W4224919704 hasConcept C165064840 @default.
- W4224919704 hasConcept C165801399 @default.
- W4224919704 hasConcept C2777402240 @default.
- W4224919704 hasConcept C28490314 @default.
- W4224919704 hasConcept C2944601119 @default.
- W4224919704 hasConcept C33923547 @default.
- W4224919704 hasConcept C40969351 @default.
- W4224919704 hasConcept C41008148 @default.
- W4224919704 hasConcept C50644808 @default.
- W4224919704 hasConcept C62520636 @default.
- W4224919704 hasConcept C66322947 @default.
- W4224919704 hasConceptScore W4224919704C105795698 @default.
- W4224919704 hasConceptScore W4224919704C111919701 @default.
- W4224919704 hasConceptScore W4224919704C118505674 @default.
- W4224919704 hasConceptScore W4224919704C121332964 @default.
- W4224919704 hasConceptScore W4224919704C142362112 @default.
- W4224919704 hasConceptScore W4224919704C153180895 @default.
- W4224919704 hasConceptScore W4224919704C153349607 @default.
- W4224919704 hasConceptScore W4224919704C154945302 @default.
- W4224919704 hasConceptScore W4224919704C165064840 @default.
- W4224919704 hasConceptScore W4224919704C165801399 @default.
- W4224919704 hasConceptScore W4224919704C2777402240 @default.
- W4224919704 hasConceptScore W4224919704C28490314 @default.
- W4224919704 hasConceptScore W4224919704C2944601119 @default.
- W4224919704 hasConceptScore W4224919704C33923547 @default.
- W4224919704 hasConceptScore W4224919704C40969351 @default.
- W4224919704 hasConceptScore W4224919704C41008148 @default.
- W4224919704 hasConceptScore W4224919704C50644808 @default.
- W4224919704 hasConceptScore W4224919704C62520636 @default.
- W4224919704 hasConceptScore W4224919704C66322947 @default.
- W4224919704 hasLocation W42249197041 @default.
- W4224919704 hasOpenAccess W4224919704 @default.
- W4224919704 hasPrimaryLocation W42249197041 @default.
- W4224919704 hasRelatedWork W13270548 @default.
- W4224919704 hasRelatedWork W13463511 @default.
- W4224919704 hasRelatedWork W2308727 @default.
- W4224919704 hasRelatedWork W4883666 @default.
- W4224919704 hasRelatedWork W5769393 @default.
- W4224919704 hasRelatedWork W7296353 @default.
- W4224919704 hasRelatedWork W7575313 @default.
- W4224919704 hasRelatedWork W813000 @default.
- W4224919704 hasRelatedWork W816280 @default.
- W4224919704 hasRelatedWork W8738421 @default.
- W4224919704 isParatext "false" @default.
- W4224919704 isRetracted "false" @default.
- W4224919704 workType "article" @default.