Matches in SemOpenAlex for { <https://semopenalex.org/work/W3201703290> ?p ?o ?g. }
- W3201703290 abstract "There has been a recent surge of interest in cross-modal pre-training. However, existed approaches pre-train a one-stream model to learn joint vision-language representation, which suffers from calculation explosion when conducting cross-modal retrieval. In this work, we propose the Contrastive Cross-Modal Knowledge Sharing Pretraining (COOKIE) method to learn universal text-image representations. There are two key designs in it, one is the weight-sharing transformer on top of the visual and textual encoders to align text and image semantically, the other is three kinds of contrastive learning designed for sharing knowledge between different modalities. Cross-modal knowledge sharing greatly promotes the learning of unimodal representation. Experiments on multi-modal matching tasks including cross-modal retrieval, text matching, and image retrieval show the effectiveness and efficiency of our pre-training framework. Our COOKIE finetuned on cross-modal datasets MSCOCO, Flickr30K, and MSRVTT achieves new state-of-the-art results while using only 3/1000 inference time comparing to one-stream models. There are also 5.7% and 3.9% improvements in the task of image retrieval and text matching. Source code will be available at https://github.com/kywen1119/COOKIE." @default.
- W3201703290 created "2021-10-11" @default.
- W3201703290 creator A5004061050 @default.
- W3201703290 creator A5031212739 @default.
- W3201703290 creator A5053615892 @default.
- W3201703290 creator A5070298925 @default.
- W3201703290 creator A5072350518 @default.
- W3201703290 creator A5086682667 @default.
- W3201703290 date "2021-10-01" @default.
- W3201703290 modified "2023-10-18" @default.
- W3201703290 title "COOKIE: Contrastive Cross-Modal Knowledge Sharing Pre-training for Vision-Language Representation" @default.
- W3201703290 cites W1773149199 @default.
- W3201703290 cites W1778065289 @default.
- W3201703290 cites W1895577753 @default.
- W3201703290 cites W1905882502 @default.
- W3201703290 cites W1912570122 @default.
- W3201703290 cites W1933349210 @default.
- W3201703290 cites W2007972815 @default.
- W3201703290 cites W2108598243 @default.
- W3201703290 cites W2194775991 @default.
- W3201703290 cites W2425121537 @default.
- W3201703290 cites W2560730294 @default.
- W3201703290 cites W2745461083 @default.
- W3201703290 cites W2798834175 @default.
- W3201703290 cites W2886641317 @default.
- W3201703290 cites W2963389687 @default.
- W3201703290 cites W2963518342 @default.
- W3201703290 cites W2964042428 @default.
- W3201703290 cites W2964280870 @default.
- W3201703290 cites W2970641574 @default.
- W3201703290 cites W2988823324 @default.
- W3201703290 cites W2998356391 @default.
- W3201703290 cites W3034239448 @default.
- W3201703290 cites W3035356601 @default.
- W3201703290 cites W3099303748 @default.
- W3201703290 cites W3104033643 @default.
- W3201703290 cites W3108008695 @default.
- W3201703290 cites W3117993946 @default.
- W3201703290 cites W3171007011 @default.
- W3201703290 cites W3174525637 @default.
- W3201703290 cites W3175888430 @default.
- W3201703290 doi "https://doi.org/10.1109/iccv48922.2021.00221" @default.
- W3201703290 hasPublicationYear "2021" @default.
- W3201703290 type Work @default.
- W3201703290 sameAs 3201703290 @default.
- W3201703290 citedByCount "12" @default.
- W3201703290 countsByYear W32017032902021 @default.
- W3201703290 countsByYear W32017032902022 @default.
- W3201703290 countsByYear W32017032902023 @default.
- W3201703290 crossrefType "proceedings-article" @default.
- W3201703290 hasAuthorship W3201703290A5004061050 @default.
- W3201703290 hasAuthorship W3201703290A5031212739 @default.
- W3201703290 hasAuthorship W3201703290A5053615892 @default.
- W3201703290 hasAuthorship W3201703290A5070298925 @default.
- W3201703290 hasAuthorship W3201703290A5072350518 @default.
- W3201703290 hasAuthorship W3201703290A5086682667 @default.
- W3201703290 hasConcept C105795698 @default.
- W3201703290 hasConcept C121332964 @default.
- W3201703290 hasConcept C154945302 @default.
- W3201703290 hasConcept C165064840 @default.
- W3201703290 hasConcept C165801399 @default.
- W3201703290 hasConcept C17744445 @default.
- W3201703290 hasConcept C185592680 @default.
- W3201703290 hasConcept C188027245 @default.
- W3201703290 hasConcept C199539241 @default.
- W3201703290 hasConcept C204321447 @default.
- W3201703290 hasConcept C23123220 @default.
- W3201703290 hasConcept C2776214188 @default.
- W3201703290 hasConcept C2776359362 @default.
- W3201703290 hasConcept C28490314 @default.
- W3201703290 hasConcept C33923547 @default.
- W3201703290 hasConcept C41008148 @default.
- W3201703290 hasConcept C59404180 @default.
- W3201703290 hasConcept C62520636 @default.
- W3201703290 hasConcept C66322947 @default.
- W3201703290 hasConcept C71139939 @default.
- W3201703290 hasConcept C94625758 @default.
- W3201703290 hasConceptScore W3201703290C105795698 @default.
- W3201703290 hasConceptScore W3201703290C121332964 @default.
- W3201703290 hasConceptScore W3201703290C154945302 @default.
- W3201703290 hasConceptScore W3201703290C165064840 @default.
- W3201703290 hasConceptScore W3201703290C165801399 @default.
- W3201703290 hasConceptScore W3201703290C17744445 @default.
- W3201703290 hasConceptScore W3201703290C185592680 @default.
- W3201703290 hasConceptScore W3201703290C188027245 @default.
- W3201703290 hasConceptScore W3201703290C199539241 @default.
- W3201703290 hasConceptScore W3201703290C204321447 @default.
- W3201703290 hasConceptScore W3201703290C23123220 @default.
- W3201703290 hasConceptScore W3201703290C2776214188 @default.
- W3201703290 hasConceptScore W3201703290C2776359362 @default.
- W3201703290 hasConceptScore W3201703290C28490314 @default.
- W3201703290 hasConceptScore W3201703290C33923547 @default.
- W3201703290 hasConceptScore W3201703290C41008148 @default.
- W3201703290 hasConceptScore W3201703290C59404180 @default.
- W3201703290 hasConceptScore W3201703290C62520636 @default.
- W3201703290 hasConceptScore W3201703290C66322947 @default.
- W3201703290 hasConceptScore W3201703290C71139939 @default.
- W3201703290 hasConceptScore W3201703290C94625758 @default.
- W3201703290 hasLocation W32017032901 @default.
- W3201703290 hasOpenAccess W3201703290 @default.