Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384648229> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4384648229 abstract "Vision Transformer (ViT) based Vision-Language Pre-training (VLP) models have demonstrated impressive performance in various tasks. However, the lengthy visual token sequences fed into ViT can lead to training inefficiency and ineffectiveness. Existing efforts address the challenge by either bottom-level patch extraction in the ViT backbone or top-level patch abstraction outside, not balancing training efficiency and effectiveness well. Inspired by text summarization in natural language processing, we propose a Bottom-Up Patch Summarization approach named BUS, coordinating bottom-level extraction and top-level abstraction to learn a concise summary of lengthy visual token sequences efficiently. Specifically, We incorporate a Text-Semantics-Aware Patch Selector (TSPS) into the ViT backbone to perform a coarse-grained visual token extraction and then attach a flexible Transformer-based Patch Abstraction Decoder (PAD) upon the backbone for top-level visual abstraction. This bottom-up collaboration enables our BUS to yield high training efficiency while maintaining or even improving effectiveness. We evaluate our approach on various visual-language understanding and generation tasks and show competitive downstream task performance while boosting the training efficiency by 50%. Additionally, our model achieves state-of-the-art performance on many downstream tasks by increasing input image resolution without increasing computational costs over baselines." @default.
- W4384648229 created "2023-07-19" @default.
- W4384648229 creator A5005674481 @default.
- W4384648229 creator A5005965903 @default.
- W4384648229 creator A5013145898 @default.
- W4384648229 creator A5013292507 @default.
- W4384648229 creator A5020554327 @default.
- W4384648229 creator A5047856952 @default.
- W4384648229 creator A5052508908 @default.
- W4384648229 creator A5065789784 @default.
- W4384648229 creator A5074446389 @default.
- W4384648229 creator A5081108106 @default.
- W4384648229 date "2023-07-17" @default.
- W4384648229 modified "2023-09-25" @default.
- W4384648229 title "BUS:Efficient and Effective Vision-language Pre-training with Bottom-Up Patch Summarization" @default.
- W4384648229 doi "https://doi.org/10.48550/arxiv.2307.08504" @default.
- W4384648229 hasPublicationYear "2023" @default.
- W4384648229 type Work @default.
- W4384648229 citedByCount "0" @default.
- W4384648229 crossrefType "posted-content" @default.
- W4384648229 hasAuthorship W4384648229A5005674481 @default.
- W4384648229 hasAuthorship W4384648229A5005965903 @default.
- W4384648229 hasAuthorship W4384648229A5013145898 @default.
- W4384648229 hasAuthorship W4384648229A5013292507 @default.
- W4384648229 hasAuthorship W4384648229A5020554327 @default.
- W4384648229 hasAuthorship W4384648229A5047856952 @default.
- W4384648229 hasAuthorship W4384648229A5052508908 @default.
- W4384648229 hasAuthorship W4384648229A5065789784 @default.
- W4384648229 hasAuthorship W4384648229A5074446389 @default.
- W4384648229 hasAuthorship W4384648229A5081108106 @default.
- W4384648229 hasBestOaLocation W43846482291 @default.
- W4384648229 hasConcept C111472728 @default.
- W4384648229 hasConcept C119599485 @default.
- W4384648229 hasConcept C124304363 @default.
- W4384648229 hasConcept C127413603 @default.
- W4384648229 hasConcept C137293760 @default.
- W4384648229 hasConcept C138885662 @default.
- W4384648229 hasConcept C154945302 @default.
- W4384648229 hasConcept C165801399 @default.
- W4384648229 hasConcept C170858558 @default.
- W4384648229 hasConcept C31258907 @default.
- W4384648229 hasConcept C41008148 @default.
- W4384648229 hasConcept C48145219 @default.
- W4384648229 hasConcept C66322947 @default.
- W4384648229 hasConceptScore W4384648229C111472728 @default.
- W4384648229 hasConceptScore W4384648229C119599485 @default.
- W4384648229 hasConceptScore W4384648229C124304363 @default.
- W4384648229 hasConceptScore W4384648229C127413603 @default.
- W4384648229 hasConceptScore W4384648229C137293760 @default.
- W4384648229 hasConceptScore W4384648229C138885662 @default.
- W4384648229 hasConceptScore W4384648229C154945302 @default.
- W4384648229 hasConceptScore W4384648229C165801399 @default.
- W4384648229 hasConceptScore W4384648229C170858558 @default.
- W4384648229 hasConceptScore W4384648229C31258907 @default.
- W4384648229 hasConceptScore W4384648229C41008148 @default.
- W4384648229 hasConceptScore W4384648229C48145219 @default.
- W4384648229 hasConceptScore W4384648229C66322947 @default.
- W4384648229 hasLocation W43846482291 @default.
- W4384648229 hasOpenAccess W4384648229 @default.
- W4384648229 hasPrimaryLocation W43846482291 @default.
- W4384648229 hasRelatedWork W2793376154 @default.
- W4384648229 hasRelatedWork W2800703702 @default.
- W4384648229 hasRelatedWork W2947686949 @default.
- W4384648229 hasRelatedWork W3215285120 @default.
- W4384648229 hasRelatedWork W4226515437 @default.
- W4384648229 hasRelatedWork W4288113047 @default.
- W4384648229 hasRelatedWork W4309436092 @default.
- W4384648229 hasRelatedWork W4313395829 @default.
- W4384648229 hasRelatedWork W4317547544 @default.
- W4384648229 hasRelatedWork W4362451017 @default.
- W4384648229 isParatext "false" @default.
- W4384648229 isRetracted "false" @default.
- W4384648229 workType "article" @default.