Matches in SemOpenAlex for { <https://semopenalex.org/work/W4319862456> ?p ?o ?g. }
Showing items 1 to 84 of
84
with 100 items per page.
- W4319862456 abstract "Self-supervised learning (SSL) has proven vital in speech and audio-related applications. The paradigm trains a general model on unlabeled data that can later be used to solve specific downstream tasks. This type of model is costly to train as it requires manipulating long input sequences that can only be handled by powerful centralised servers. Surprisingly, despite many attempts to increase training efficiency through model compression, the effects of truncating input sequence lengths to reduce computation have not been studied. In this paper, we provide the first empirical study of SSL pre-training for different specified sequence lengths and link this to various downstream tasks. We find that training on short sequences can dramatically reduce resource costs while retaining a satisfactory performance for all tasks. This simple one-line change would promote the migration of SSL training from data centres to user-end edge devices for more realistic and personalised applications." @default.
- W4319862456 created "2023-02-11" @default.
- W4319862456 creator A5031116318 @default.
- W4319862456 creator A5045638679 @default.
- W4319862456 creator A5077645636 @default.
- W4319862456 creator A5082863098 @default.
- W4319862456 creator A5089505434 @default.
- W4319862456 date "2023-01-09" @default.
- W4319862456 modified "2023-09-27" @default.
- W4319862456 title "Match to Win: Analysing Sequences Lengths for Efficient Self-Supervised Learning in Speech and Audio" @default.
- W4319862456 cites W1494198834 @default.
- W4319862456 cites W2146334809 @default.
- W4319862456 cites W2526425061 @default.
- W4319862456 cites W2890964092 @default.
- W4319862456 cites W2892009249 @default.
- W4319862456 cites W2979826702 @default.
- W4319862456 cites W2981087920 @default.
- W4319862456 cites W3015213852 @default.
- W4319862456 cites W3148040514 @default.
- W4319862456 cites W3197580070 @default.
- W4319862456 cites W3209059054 @default.
- W4319862456 cites W4296070200 @default.
- W4319862456 doi "https://doi.org/10.1109/slt54892.2023.10023410" @default.
- W4319862456 hasPublicationYear "2023" @default.
- W4319862456 type Work @default.
- W4319862456 citedByCount "0" @default.
- W4319862456 crossrefType "proceedings-article" @default.
- W4319862456 hasAuthorship W4319862456A5031116318 @default.
- W4319862456 hasAuthorship W4319862456A5045638679 @default.
- W4319862456 hasAuthorship W4319862456A5077645636 @default.
- W4319862456 hasAuthorship W4319862456A5082863098 @default.
- W4319862456 hasAuthorship W4319862456A5089505434 @default.
- W4319862456 hasConcept C111919701 @default.
- W4319862456 hasConcept C11413529 @default.
- W4319862456 hasConcept C119857082 @default.
- W4319862456 hasConcept C138236772 @default.
- W4319862456 hasConcept C154945302 @default.
- W4319862456 hasConcept C162307627 @default.
- W4319862456 hasConcept C190839683 @default.
- W4319862456 hasConcept C205649164 @default.
- W4319862456 hasConcept C2778112365 @default.
- W4319862456 hasConcept C28490314 @default.
- W4319862456 hasConcept C31258907 @default.
- W4319862456 hasConcept C41008148 @default.
- W4319862456 hasConcept C45374587 @default.
- W4319862456 hasConcept C54355233 @default.
- W4319862456 hasConcept C58640448 @default.
- W4319862456 hasConcept C79974875 @default.
- W4319862456 hasConcept C86803240 @default.
- W4319862456 hasConcept C93996380 @default.
- W4319862456 hasConceptScore W4319862456C111919701 @default.
- W4319862456 hasConceptScore W4319862456C11413529 @default.
- W4319862456 hasConceptScore W4319862456C119857082 @default.
- W4319862456 hasConceptScore W4319862456C138236772 @default.
- W4319862456 hasConceptScore W4319862456C154945302 @default.
- W4319862456 hasConceptScore W4319862456C162307627 @default.
- W4319862456 hasConceptScore W4319862456C190839683 @default.
- W4319862456 hasConceptScore W4319862456C205649164 @default.
- W4319862456 hasConceptScore W4319862456C2778112365 @default.
- W4319862456 hasConceptScore W4319862456C28490314 @default.
- W4319862456 hasConceptScore W4319862456C31258907 @default.
- W4319862456 hasConceptScore W4319862456C41008148 @default.
- W4319862456 hasConceptScore W4319862456C45374587 @default.
- W4319862456 hasConceptScore W4319862456C54355233 @default.
- W4319862456 hasConceptScore W4319862456C58640448 @default.
- W4319862456 hasConceptScore W4319862456C79974875 @default.
- W4319862456 hasConceptScore W4319862456C86803240 @default.
- W4319862456 hasConceptScore W4319862456C93996380 @default.
- W4319862456 hasLocation W43198624561 @default.
- W4319862456 hasOpenAccess W4319862456 @default.
- W4319862456 hasPrimaryLocation W43198624561 @default.
- W4319862456 hasRelatedWork W2989221764 @default.
- W4319862456 hasRelatedWork W3042492850 @default.
- W4319862456 hasRelatedWork W3106610906 @default.
- W4319862456 hasRelatedWork W3106904549 @default.
- W4319862456 hasRelatedWork W3185724185 @default.
- W4319862456 hasRelatedWork W3191281725 @default.
- W4319862456 hasRelatedWork W4205505298 @default.
- W4319862456 hasRelatedWork W4292394195 @default.
- W4319862456 hasRelatedWork W4311454612 @default.
- W4319862456 hasRelatedWork W4316042121 @default.
- W4319862456 isParatext "false" @default.
- W4319862456 isRetracted "false" @default.
- W4319862456 workType "article" @default.