Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385572899> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4385572899 abstract "Vision Transformers (ViTs) have been widely used in large-scale Vision and Language Pre-training (VLP) models. Though previous VLP works have proved the effectiveness of ViTs, they still suffer from computational efficiency brought by the long visual sequence. To tackle this problem, in this paper, we propose an efficient vision-and-language pre-training model with Text-Relevant Image Patch Selection, namely TRIPS, which reduces the visual sequence progressively with a text-guided patch-selection layer in the visual backbone for efficient training and inference. The patch-selection layer can dynamically compute text-dependent visual attention to identify the attentive image tokens with text guidance and fuse inattentive ones in an end-to-end manner. Meanwhile, TRIPS does not introduce extra parameters to ViTs. Experimental results on a variety of popular benchmark datasets demonstrate that TRIPS gain a speedup of 40% over previous similar VLP models, yet with competitive or better downstream task performance." @default.
- W4385572899 created "2023-08-05" @default.
- W4385572899 creator A5005674481 @default.
- W4385572899 creator A5013145898 @default.
- W4385572899 creator A5020554327 @default.
- W4385572899 creator A5047856952 @default.
- W4385572899 creator A5064249897 @default.
- W4385572899 creator A5065789784 @default.
- W4385572899 creator A5081108106 @default.
- W4385572899 creator A5084741576 @default.
- W4385572899 date "2022-01-01" @default.
- W4385572899 modified "2023-09-25" @default.
- W4385572899 title "TRIPS: Efficient Vision-and-Language Pre-training with Text-Relevant Image Patch Selection" @default.
- W4385572899 doi "https://doi.org/10.18653/v1/2022.emnlp-main.273" @default.
- W4385572899 hasPublicationYear "2022" @default.
- W4385572899 type Work @default.
- W4385572899 citedByCount "0" @default.
- W4385572899 crossrefType "proceedings-article" @default.
- W4385572899 hasAuthorship W4385572899A5005674481 @default.
- W4385572899 hasAuthorship W4385572899A5013145898 @default.
- W4385572899 hasAuthorship W4385572899A5020554327 @default.
- W4385572899 hasAuthorship W4385572899A5047856952 @default.
- W4385572899 hasAuthorship W4385572899A5064249897 @default.
- W4385572899 hasAuthorship W4385572899A5065789784 @default.
- W4385572899 hasAuthorship W4385572899A5081108106 @default.
- W4385572899 hasAuthorship W4385572899A5084741576 @default.
- W4385572899 hasBestOaLocation W43855728991 @default.
- W4385572899 hasConcept C111919701 @default.
- W4385572899 hasConcept C119857082 @default.
- W4385572899 hasConcept C13280743 @default.
- W4385572899 hasConcept C137293760 @default.
- W4385572899 hasConcept C154945302 @default.
- W4385572899 hasConcept C157085824 @default.
- W4385572899 hasConcept C173608175 @default.
- W4385572899 hasConcept C185798385 @default.
- W4385572899 hasConcept C204321447 @default.
- W4385572899 hasConcept C205649164 @default.
- W4385572899 hasConcept C2776214188 @default.
- W4385572899 hasConcept C31972630 @default.
- W4385572899 hasConcept C41008148 @default.
- W4385572899 hasConcept C68339613 @default.
- W4385572899 hasConcept C81917197 @default.
- W4385572899 hasConceptScore W4385572899C111919701 @default.
- W4385572899 hasConceptScore W4385572899C119857082 @default.
- W4385572899 hasConceptScore W4385572899C13280743 @default.
- W4385572899 hasConceptScore W4385572899C137293760 @default.
- W4385572899 hasConceptScore W4385572899C154945302 @default.
- W4385572899 hasConceptScore W4385572899C157085824 @default.
- W4385572899 hasConceptScore W4385572899C173608175 @default.
- W4385572899 hasConceptScore W4385572899C185798385 @default.
- W4385572899 hasConceptScore W4385572899C204321447 @default.
- W4385572899 hasConceptScore W4385572899C205649164 @default.
- W4385572899 hasConceptScore W4385572899C2776214188 @default.
- W4385572899 hasConceptScore W4385572899C31972630 @default.
- W4385572899 hasConceptScore W4385572899C41008148 @default.
- W4385572899 hasConceptScore W4385572899C68339613 @default.
- W4385572899 hasConceptScore W4385572899C81917197 @default.
- W4385572899 hasLocation W43855728991 @default.
- W4385572899 hasOpenAccess W4385572899 @default.
- W4385572899 hasPrimaryLocation W43855728991 @default.
- W4385572899 hasRelatedWork W2359001871 @default.
- W4385572899 hasRelatedWork W2949992439 @default.
- W4385572899 hasRelatedWork W2972987451 @default.
- W4385572899 hasRelatedWork W3008756425 @default.
- W4385572899 hasRelatedWork W3014568172 @default.
- W4385572899 hasRelatedWork W3035030897 @default.
- W4385572899 hasRelatedWork W3104738015 @default.
- W4385572899 hasRelatedWork W4285077633 @default.
- W4385572899 hasRelatedWork W4320486724 @default.
- W4385572899 hasRelatedWork W1506942559 @default.
- W4385572899 isParatext "false" @default.
- W4385572899 isRetracted "false" @default.
- W4385572899 workType "article" @default.