Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385815508> ?p ?o ?g. }
- W4385815508 abstract "With the large-scale video-text datasets being collected, learning general visual-textual representation has gained increasing attention. While recent methods are designed with the assumption that the alt-text description naturally conveys the meaning and context of the video in semantics (i.e. well aligned with each other), it is unlikely to be satisfied for the Internet data, which potentially harms the quality of the learned visual-textual representation. To address this challenge, we first revisit three mainstream approaches: correspondence modeling, contrastive learning and predictive coding, demonstrating that a simple co-training strategy with these methods leads to a clear improvement in performance. To further explore the complementary nature of different training strategies, we propose a simple yet effective joint training framework that factorizes the total objective into conditional ones, termed as Cali-NCE <sup xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>1</sup> . Our method first estimates confidence scores for measuring the correspondence between video and text descriptions, and the scores are later used to calibrate the sample weightings during contrastive training. Through extensive experiments, we show that the proposed approach achieves state-of-the-art performance on multiple downstream tasks: text-to-video retrieval, video action recognition, and video retrieval." @default.
- W4385815508 created "2023-08-15" @default.
- W4385815508 creator A5010087030 @default.
- W4385815508 creator A5017599481 @default.
- W4385815508 creator A5072341936 @default.
- W4385815508 creator A5076097168 @default.
- W4385815508 date "2023-06-01" @default.
- W4385815508 modified "2023-09-26" @default.
- W4385815508 title "Cali-NCE: Boosting Cross-modal Video Representation Learning with Calibrated Alignment" @default.
- W4385815508 cites W1573040851 @default.
- W4385815508 cites W1933349210 @default.
- W4385815508 cites W1957706851 @default.
- W4385815508 cites W1976849453 @default.
- W4385815508 cites W2077947298 @default.
- W4385815508 cites W2087305065 @default.
- W4385815508 cites W2126579184 @default.
- W4385815508 cites W2302086703 @default.
- W4385815508 cites W2425121537 @default.
- W4385815508 cites W2463955103 @default.
- W4385815508 cites W2550462002 @default.
- W4385815508 cites W2565656701 @default.
- W4385815508 cites W2606473278 @default.
- W4385815508 cites W2798271879 @default.
- W4385815508 cites W2798991696 @default.
- W4385815508 cites W2799087757 @default.
- W4385815508 cites W2808399042 @default.
- W4385815508 cites W2948242301 @default.
- W4385815508 cites W2963389687 @default.
- W4385815508 cites W2963541336 @default.
- W4385815508 cites W2963758027 @default.
- W4385815508 cites W2963814513 @default.
- W4385815508 cites W2964037671 @default.
- W4385815508 cites W2981851019 @default.
- W4385815508 cites W2984008963 @default.
- W4385815508 cites W3010874390 @default.
- W4385815508 cites W3035265375 @default.
- W4385815508 cites W3035524453 @default.
- W4385815508 cites W3035635319 @default.
- W4385815508 cites W3110190397 @default.
- W4385815508 cites W3130796238 @default.
- W4385815508 cites W3145385912 @default.
- W4385815508 cites W3158986867 @default.
- W4385815508 cites W3168640669 @default.
- W4385815508 cites W3204588463 @default.
- W4385815508 cites W4214507759 @default.
- W4385815508 cites W4312271977 @default.
- W4385815508 cites W4313186260 @default.
- W4385815508 doi "https://doi.org/10.1109/cvprw59228.2023.00672" @default.
- W4385815508 hasPublicationYear "2023" @default.
- W4385815508 type Work @default.
- W4385815508 citedByCount "0" @default.
- W4385815508 crossrefType "proceedings-article" @default.
- W4385815508 hasAuthorship W4385815508A5010087030 @default.
- W4385815508 hasAuthorship W4385815508A5017599481 @default.
- W4385815508 hasAuthorship W4385815508A5072341936 @default.
- W4385815508 hasAuthorship W4385815508A5076097168 @default.
- W4385815508 hasConcept C110875604 @default.
- W4385815508 hasConcept C136764020 @default.
- W4385815508 hasConcept C151730666 @default.
- W4385815508 hasConcept C154945302 @default.
- W4385815508 hasConcept C17744445 @default.
- W4385815508 hasConcept C184337299 @default.
- W4385815508 hasConcept C199360897 @default.
- W4385815508 hasConcept C199539241 @default.
- W4385815508 hasConcept C204321447 @default.
- W4385815508 hasConcept C23123220 @default.
- W4385815508 hasConcept C2776359362 @default.
- W4385815508 hasConcept C2779343474 @default.
- W4385815508 hasConcept C2779789524 @default.
- W4385815508 hasConcept C41008148 @default.
- W4385815508 hasConcept C46686674 @default.
- W4385815508 hasConcept C59404180 @default.
- W4385815508 hasConcept C86803240 @default.
- W4385815508 hasConcept C94625758 @default.
- W4385815508 hasConceptScore W4385815508C110875604 @default.
- W4385815508 hasConceptScore W4385815508C136764020 @default.
- W4385815508 hasConceptScore W4385815508C151730666 @default.
- W4385815508 hasConceptScore W4385815508C154945302 @default.
- W4385815508 hasConceptScore W4385815508C17744445 @default.
- W4385815508 hasConceptScore W4385815508C184337299 @default.
- W4385815508 hasConceptScore W4385815508C199360897 @default.
- W4385815508 hasConceptScore W4385815508C199539241 @default.
- W4385815508 hasConceptScore W4385815508C204321447 @default.
- W4385815508 hasConceptScore W4385815508C23123220 @default.
- W4385815508 hasConceptScore W4385815508C2776359362 @default.
- W4385815508 hasConceptScore W4385815508C2779343474 @default.
- W4385815508 hasConceptScore W4385815508C2779789524 @default.
- W4385815508 hasConceptScore W4385815508C41008148 @default.
- W4385815508 hasConceptScore W4385815508C46686674 @default.
- W4385815508 hasConceptScore W4385815508C59404180 @default.
- W4385815508 hasConceptScore W4385815508C86803240 @default.
- W4385815508 hasConceptScore W4385815508C94625758 @default.
- W4385815508 hasLocation W43858155081 @default.
- W4385815508 hasOpenAccess W4385815508 @default.
- W4385815508 hasPrimaryLocation W43858155081 @default.
- W4385815508 hasRelatedWork W1998940060 @default.
- W4385815508 hasRelatedWork W2086064646 @default.
- W4385815508 hasRelatedWork W2115485936 @default.
- W4385815508 hasRelatedWork W2119135658 @default.
- W4385815508 hasRelatedWork W2293457016 @default.