Matches in SemOpenAlex for { <https://semopenalex.org/work/W4225605140> ?p ?o ?g. }
Showing items 1 to 82 of
82
with 100 items per page.
- W4225605140 endingPage "442" @default.
- W4225605140 startingPage "431" @default.
- W4225605140 abstract "AbstractImage-Text Retrieval (ITR) enables users to retrieve relevant contents from different modalities and has attracted considerable attention. Existing approaches typically utilize contrastive loss functions to conduct contrastive learning in the common embedding space, where they aim at pulling semantically related pairs closer while pushing away unrelated pairs. However, we argue that this behaviour is too strict: these approaches neglect to address the inherent misalignments from potential semantically related samples. For example, it commonly exists more than one positive samples in the current batch for a given query and previous methods enforce them apart even if they are semantically related, which leads to a sub-optimal and contradictory optimization direction and then decreases the retrieval performance. In this paper, a Multiple Positives Enhanced Noise Contrastive Estimation learning objective is proposed to alleviate the diversion noise by leveraging and optimizing multiple positive pairs overall for each sample in a mini-batch. We demonstrate the effectiveness of our approach on MS-COCO and Flickr30K datasets for image-to-text and text-to-image retrieval.KeywordsImage-text retrievalContrastive learningNoise contrastive estimation" @default.
- W4225605140 created "2022-05-05" @default.
- W4225605140 creator A5013692772 @default.
- W4225605140 creator A5056706680 @default.
- W4225605140 creator A5064842058 @default.
- W4225605140 date "2022-01-01" @default.
- W4225605140 modified "2023-09-27" @default.
- W4225605140 title "A Multiple Positives Enhanced NCE Loss for Image-Text Retrieval" @default.
- W4225605140 cites W1566289585 @default.
- W4225605140 cites W1861492603 @default.
- W4225605140 cites W1905882502 @default.
- W4225605140 cites W1957706851 @default.
- W4225605140 cites W2185175083 @default.
- W4225605140 cites W2745461083 @default.
- W4225605140 cites W2778100917 @default.
- W4225605140 cites W2962964995 @default.
- W4225605140 cites W2963040148 @default.
- W4225605140 cites W2963467339 @default.
- W4225605140 cites W2964120214 @default.
- W4225605140 cites W2981851019 @default.
- W4225605140 cites W2985951359 @default.
- W4225605140 cites W2988823324 @default.
- W4225605140 cites W3028452400 @default.
- W4225605140 cites W3035524453 @default.
- W4225605140 cites W3091588028 @default.
- W4225605140 cites W3092820619 @default.
- W4225605140 cites W3156636935 @default.
- W4225605140 cites W3160179442 @default.
- W4225605140 cites W3213100861 @default.
- W4225605140 cites W3008356057 @default.
- W4225605140 doi "https://doi.org/10.1007/978-3-030-98358-1_34" @default.
- W4225605140 hasPublicationYear "2022" @default.
- W4225605140 type Work @default.
- W4225605140 citedByCount "0" @default.
- W4225605140 crossrefType "book-chapter" @default.
- W4225605140 hasAuthorship W4225605140A5013692772 @default.
- W4225605140 hasAuthorship W4225605140A5056706680 @default.
- W4225605140 hasAuthorship W4225605140A5064842058 @default.
- W4225605140 hasConcept C115961682 @default.
- W4225605140 hasConcept C153180895 @default.
- W4225605140 hasConcept C154945302 @default.
- W4225605140 hasConcept C1667742 @default.
- W4225605140 hasConcept C185592680 @default.
- W4225605140 hasConcept C198531522 @default.
- W4225605140 hasConcept C204321447 @default.
- W4225605140 hasConcept C23123220 @default.
- W4225605140 hasConcept C41008148 @default.
- W4225605140 hasConcept C41608201 @default.
- W4225605140 hasConcept C43617362 @default.
- W4225605140 hasConcept C64869954 @default.
- W4225605140 hasConcept C99498987 @default.
- W4225605140 hasConceptScore W4225605140C115961682 @default.
- W4225605140 hasConceptScore W4225605140C153180895 @default.
- W4225605140 hasConceptScore W4225605140C154945302 @default.
- W4225605140 hasConceptScore W4225605140C1667742 @default.
- W4225605140 hasConceptScore W4225605140C185592680 @default.
- W4225605140 hasConceptScore W4225605140C198531522 @default.
- W4225605140 hasConceptScore W4225605140C204321447 @default.
- W4225605140 hasConceptScore W4225605140C23123220 @default.
- W4225605140 hasConceptScore W4225605140C41008148 @default.
- W4225605140 hasConceptScore W4225605140C41608201 @default.
- W4225605140 hasConceptScore W4225605140C43617362 @default.
- W4225605140 hasConceptScore W4225605140C64869954 @default.
- W4225605140 hasConceptScore W4225605140C99498987 @default.
- W4225605140 hasLocation W42256051401 @default.
- W4225605140 hasOpenAccess W4225605140 @default.
- W4225605140 hasPrimaryLocation W42256051401 @default.
- W4225605140 hasRelatedWork W1489668150 @default.
- W4225605140 hasRelatedWork W1554359492 @default.
- W4225605140 hasRelatedWork W2106795979 @default.
- W4225605140 hasRelatedWork W2130538965 @default.
- W4225605140 hasRelatedWork W2150769968 @default.
- W4225605140 hasRelatedWork W3012978760 @default.
- W4225605140 hasRelatedWork W3086857729 @default.
- W4225605140 hasRelatedWork W3107474891 @default.
- W4225605140 hasRelatedWork W4211210873 @default.
- W4225605140 hasRelatedWork W98013039 @default.
- W4225605140 isParatext "false" @default.
- W4225605140 isRetracted "false" @default.
- W4225605140 workType "book-chapter" @default.