Matches in SemOpenAlex for { <https://semopenalex.org/work/W4226380987> ?p ?o ?g. }
- W4226380987 abstract "Recently proposed self-supervised learning approaches have been successful for pre-training speech representation models. The utility of these learned representations has been observed empirically, but not much has been studied about the type or extent of information encoded in the pre-trained representations themselves. Developing such insights can help understand the capabilities and limits of these models and enable the research community to more efficiently develop their usage for downstream applications. In this work, we begin to fill this gap by examining one recent and successful pre-trained model (wav2vec 2.0), via its intermediate representation vectors, using a suite of analysis tools. We use the metrics of canonical correlation, mutual information, and performance on simple downstream tasks with non-parametric probes, in order to (i) query for acoustic and linguistic information content, (ii) characterize the evolution of information across model layers, and (iii) understand how fine-tuning the model for automatic speech recognition (ASR) affects these observations. Our findings motivate modifying the fine-tuning protocol for ASR, which produces improved word error rates in a low-resource setting." @default.
- W4226380987 created "2022-05-05" @default.
- W4226380987 creator A5015602781 @default.
- W4226380987 creator A5049062211 @default.
- W4226380987 creator A5059672669 @default.
- W4226380987 date "2021-12-13" @default.
- W4226380987 modified "2023-10-03" @default.
- W4226380987 title "Layer-Wise Analysis of a Self-Supervised Speech Representation Model" @default.
- W4226380987 cites W1494198834 @default.
- W4226380987 cites W1545920196 @default.
- W4226380987 cites W2127141656 @default.
- W4226380987 cites W2250539671 @default.
- W4226380987 cites W2251253014 @default.
- W4226380987 cites W2407151108 @default.
- W4226380987 cites W2586148577 @default.
- W4226380987 cites W2906152891 @default.
- W4226380987 cites W2932675979 @default.
- W4226380987 cites W2946417913 @default.
- W4226380987 cites W2963259843 @default.
- W4226380987 cites W2963425185 @default.
- W4226380987 cites W2970820321 @default.
- W4226380987 cites W2972584841 @default.
- W4226380987 cites W2973049979 @default.
- W4226380987 cites W2973157397 @default.
- W4226380987 cites W2982223350 @default.
- W4226380987 cites W2988217457 @default.
- W4226380987 cites W2995181338 @default.
- W4226380987 cites W3003875258 @default.
- W4226380987 cites W3015265920 @default.
- W4226380987 cites W3016011332 @default.
- W4226380987 cites W3034273309 @default.
- W4226380987 cites W3095706145 @default.
- W4226380987 cites W3096196861 @default.
- W4226380987 cites W3146777637 @default.
- W4226380987 cites W3160799772 @default.
- W4226380987 cites W3162133897 @default.
- W4226380987 cites W3197580070 @default.
- W4226380987 cites W3198299542 @default.
- W4226380987 cites W3198771897 @default.
- W4226380987 cites W343636949 @default.
- W4226380987 cites W4237723258 @default.
- W4226380987 cites W3163596720 @default.
- W4226380987 doi "https://doi.org/10.1109/asru51503.2021.9688093" @default.
- W4226380987 hasPublicationYear "2021" @default.
- W4226380987 type Work @default.
- W4226380987 citedByCount "33" @default.
- W4226380987 countsByYear W42263809872022 @default.
- W4226380987 countsByYear W42263809872023 @default.
- W4226380987 crossrefType "proceedings-article" @default.
- W4226380987 hasAuthorship W4226380987A5015602781 @default.
- W4226380987 hasAuthorship W4226380987A5049062211 @default.
- W4226380987 hasAuthorship W4226380987A5059672669 @default.
- W4226380987 hasBestOaLocation W42263809872 @default.
- W4226380987 hasConcept C105795698 @default.
- W4226380987 hasConcept C117251300 @default.
- W4226380987 hasConcept C119857082 @default.
- W4226380987 hasConcept C138885662 @default.
- W4226380987 hasConcept C152139883 @default.
- W4226380987 hasConcept C153874254 @default.
- W4226380987 hasConcept C154945302 @default.
- W4226380987 hasConcept C162324750 @default.
- W4226380987 hasConcept C166957645 @default.
- W4226380987 hasConcept C17744445 @default.
- W4226380987 hasConcept C199539241 @default.
- W4226380987 hasConcept C204321447 @default.
- W4226380987 hasConcept C21547014 @default.
- W4226380987 hasConcept C24574437 @default.
- W4226380987 hasConcept C2776207758 @default.
- W4226380987 hasConcept C2776359362 @default.
- W4226380987 hasConcept C28490314 @default.
- W4226380987 hasConcept C33923547 @default.
- W4226380987 hasConcept C41008148 @default.
- W4226380987 hasConcept C41895202 @default.
- W4226380987 hasConcept C59404180 @default.
- W4226380987 hasConcept C79581498 @default.
- W4226380987 hasConcept C90805587 @default.
- W4226380987 hasConcept C94625758 @default.
- W4226380987 hasConcept C95457728 @default.
- W4226380987 hasConceptScore W4226380987C105795698 @default.
- W4226380987 hasConceptScore W4226380987C117251300 @default.
- W4226380987 hasConceptScore W4226380987C119857082 @default.
- W4226380987 hasConceptScore W4226380987C138885662 @default.
- W4226380987 hasConceptScore W4226380987C152139883 @default.
- W4226380987 hasConceptScore W4226380987C153874254 @default.
- W4226380987 hasConceptScore W4226380987C154945302 @default.
- W4226380987 hasConceptScore W4226380987C162324750 @default.
- W4226380987 hasConceptScore W4226380987C166957645 @default.
- W4226380987 hasConceptScore W4226380987C17744445 @default.
- W4226380987 hasConceptScore W4226380987C199539241 @default.
- W4226380987 hasConceptScore W4226380987C204321447 @default.
- W4226380987 hasConceptScore W4226380987C21547014 @default.
- W4226380987 hasConceptScore W4226380987C24574437 @default.
- W4226380987 hasConceptScore W4226380987C2776207758 @default.
- W4226380987 hasConceptScore W4226380987C2776359362 @default.
- W4226380987 hasConceptScore W4226380987C28490314 @default.
- W4226380987 hasConceptScore W4226380987C33923547 @default.
- W4226380987 hasConceptScore W4226380987C41008148 @default.
- W4226380987 hasConceptScore W4226380987C41895202 @default.
- W4226380987 hasConceptScore W4226380987C59404180 @default.
- W4226380987 hasConceptScore W4226380987C79581498 @default.