Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204383925> ?p ?o ?g. }
- W3204383925 abstract "Contrastive learning allows us to flexibly define powerful losses by contrasting positive pairs from sets of negative samples. Recently, the principle has also been used to learn cross-modal embeddings for video and text, yet without exploiting its full potential. In particular, previous losses do not take the intra-modality similarities into account, which leads to inefficient embeddings, as the same content is mapped to multiple points in the embedding space. With CrossCLR, we present a contrastive loss that fixes this issue. Moreover, we define sets of highly related samples in terms of their input embeddings and exclude them from the negative samples to avoid issues with false negatives. We show that these principles consistently improve the quality of the learned embeddings. The joint embeddings learned with CrossCLR extend the state of the art in video-text retrieval on Youcook2 and LSMDC datasets and in video captioning on Youcook2 dataset by a large margin. We also demonstrate the generality of the concept by learning improved joint embeddings for other pairs of modalities." @default.
- W3204383925 created "2021-10-11" @default.
- W3204383925 creator A5019191351 @default.
- W3204383925 creator A5045033461 @default.
- W3204383925 creator A5055614518 @default.
- W3204383925 creator A5070290355 @default.
- W3204383925 date "2021-09-30" @default.
- W3204383925 modified "2023-09-27" @default.
- W3204383925 title "CrossCLR: Cross-modal Contrastive Learning For Multi-modal Video Representations" @default.
- W3204383925 cites W1956340063 @default.
- W3204383925 cites W1957706851 @default.
- W3204383925 cites W1975517671 @default.
- W3204383925 cites W2078238240 @default.
- W3204383925 cites W2101105183 @default.
- W3204383925 cites W2133459682 @default.
- W3204383925 cites W2138621090 @default.
- W3204383925 cites W2154652894 @default.
- W3204383925 cites W2565656701 @default.
- W3204383925 cites W2613718673 @default.
- W3204383925 cites W2732026016 @default.
- W3204383925 cites W2784025607 @default.
- W3204383925 cites W2796207103 @default.
- W3204383925 cites W2798991696 @default.
- W3204383925 cites W2808399042 @default.
- W3204383925 cites W2842511635 @default.
- W3204383925 cites W2883910824 @default.
- W3204383925 cites W2885775891 @default.
- W3204383925 cites W2963389687 @default.
- W3204383925 cites W2963775347 @default.
- W3204383925 cites W2981851019 @default.
- W3204383925 cites W2984008963 @default.
- W3204383925 cites W2984287396 @default.
- W3204383925 cites W2985185987 @default.
- W3204383925 cites W2990726673 @default.
- W3204383925 cites W3009561768 @default.
- W3204383925 cites W3023441976 @default.
- W3204383925 cites W3034978746 @default.
- W3204383925 cites W3035060554 @default.
- W3204383925 cites W3035237998 @default.
- W3204383925 cites W3035524453 @default.
- W3204383925 cites W3035635319 @default.
- W3204383925 cites W3036224891 @default.
- W3204383925 cites W3036982689 @default.
- W3204383925 cites W3047826509 @default.
- W3204383925 cites W3094454579 @default.
- W3204383925 cites W3100345210 @default.
- W3204383925 cites W3102363610 @default.
- W3204383925 cites W3105479157 @default.
- W3204383925 cites W3106362698 @default.
- W3204383925 cites W3106485021 @default.
- W3204383925 cites W3112449926 @default.
- W3204383925 cites W3123939835 @default.
- W3204383925 cites W3126686773 @default.
- W3204383925 cites W3129719298 @default.
- W3204383925 hasPublicationYear "2021" @default.
- W3204383925 type Work @default.
- W3204383925 sameAs 3204383925 @default.
- W3204383925 citedByCount "0" @default.
- W3204383925 crossrefType "posted-content" @default.
- W3204383925 hasAuthorship W3204383925A5019191351 @default.
- W3204383925 hasAuthorship W3204383925A5045033461 @default.
- W3204383925 hasAuthorship W3204383925A5055614518 @default.
- W3204383925 hasAuthorship W3204383925A5070290355 @default.
- W3204383925 hasConcept C111472728 @default.
- W3204383925 hasConcept C111919701 @default.
- W3204383925 hasConcept C115961682 @default.
- W3204383925 hasConcept C119857082 @default.
- W3204383925 hasConcept C127413603 @default.
- W3204383925 hasConcept C138885662 @default.
- W3204383925 hasConcept C144024400 @default.
- W3204383925 hasConcept C154945302 @default.
- W3204383925 hasConcept C15744967 @default.
- W3204383925 hasConcept C157657479 @default.
- W3204383925 hasConcept C170154142 @default.
- W3204383925 hasConcept C18555067 @default.
- W3204383925 hasConcept C185592680 @default.
- W3204383925 hasConcept C188027245 @default.
- W3204383925 hasConcept C204321447 @default.
- W3204383925 hasConcept C2778572836 @default.
- W3204383925 hasConcept C2779530757 @default.
- W3204383925 hasConcept C2779903281 @default.
- W3204383925 hasConcept C2780226545 @default.
- W3204383925 hasConcept C2780767217 @default.
- W3204383925 hasConcept C36289849 @default.
- W3204383925 hasConcept C41008148 @default.
- W3204383925 hasConcept C41608201 @default.
- W3204383925 hasConcept C542102704 @default.
- W3204383925 hasConcept C71139939 @default.
- W3204383925 hasConcept C774472 @default.
- W3204383925 hasConceptScore W3204383925C111472728 @default.
- W3204383925 hasConceptScore W3204383925C111919701 @default.
- W3204383925 hasConceptScore W3204383925C115961682 @default.
- W3204383925 hasConceptScore W3204383925C119857082 @default.
- W3204383925 hasConceptScore W3204383925C127413603 @default.
- W3204383925 hasConceptScore W3204383925C138885662 @default.
- W3204383925 hasConceptScore W3204383925C144024400 @default.
- W3204383925 hasConceptScore W3204383925C154945302 @default.
- W3204383925 hasConceptScore W3204383925C15744967 @default.
- W3204383925 hasConceptScore W3204383925C157657479 @default.
- W3204383925 hasConceptScore W3204383925C170154142 @default.