Matches in SemOpenAlex for { <https://semopenalex.org/work/W3170554424> ?p ?o ?g. }
- W3170554424 abstract "Self-supervised learning has recently shown great potential in vision tasks through contrastive learning, which aims to discriminate each image, or instance, in the dataset. However, such instance-level learning ignores the semantic relationship among instances and sometimes undesirably repels the anchor from the semantically similar samples, termed as false negatives. In this work, we show that the unfavorable effect from false negatives is more significant for the large-scale datasets with more semantic concepts. To address the issue, we propose a novel self-supervised contrastive learning framework that incrementally detects and explicitly removes the false negative samples. Specifically, following the training process, our method dynamically detects increasing high-quality false negatives considering that the encoder gradually improves and the embedding space becomes more semantically structural. Next, we discuss two strategies to explicitly remove the detected false negatives during contrastive learning. Extensive experiments show that our framework outperforms other self-supervised contrastive learning methods on multiple benchmarks in a limited resource setup." @default.
- W3170554424 created "2021-06-22" @default.
- W3170554424 creator A5009305504 @default.
- W3170554424 creator A5058383656 @default.
- W3170554424 creator A5063179713 @default.
- W3170554424 creator A5064935846 @default.
- W3170554424 creator A5090586839 @default.
- W3170554424 date "2021-06-07" @default.
- W3170554424 modified "2023-09-25" @default.
- W3170554424 title "Incremental False Negative Detection for Contrastive Learning" @default.
- W3170554424 cites W1861492603 @default.
- W3170554424 cites W1998871699 @default.
- W3170554424 cites W2009797711 @default.
- W3170554424 cites W2031489346 @default.
- W3170554424 cites W2034767282 @default.
- W3170554424 cites W2038721957 @default.
- W3170554424 cites W2049017883 @default.
- W3170554424 cites W2097645701 @default.
- W3170554424 cites W2108598243 @default.
- W3170554424 cites W2134670479 @default.
- W3170554424 cites W2144796873 @default.
- W3170554424 cites W2152790380 @default.
- W3170554424 cites W2163605009 @default.
- W3170554424 cites W2187089797 @default.
- W3170554424 cites W2194775991 @default.
- W3170554424 cites W2321533354 @default.
- W3170554424 cites W2518108298 @default.
- W3170554424 cites W2606611007 @default.
- W3170554424 cites W2842511635 @default.
- W3170554424 cites W2883725317 @default.
- W3170554424 cites W2887997457 @default.
- W3170554424 cites W2917551568 @default.
- W3170554424 cites W2949517790 @default.
- W3170554424 cites W2962742544 @default.
- W3170554424 cites W2962852342 @default.
- W3170554424 cites W2963026686 @default.
- W3170554424 cites W2963150697 @default.
- W3170554424 cites W2963350250 @default.
- W3170554424 cites W2963420272 @default.
- W3170554424 cites W2964074409 @default.
- W3170554424 cites W2979579363 @default.
- W3170554424 cites W2987741655 @default.
- W3170554424 cites W2995181141 @default.
- W3170554424 cites W2998388430 @default.
- W3170554424 cites W2998702515 @default.
- W3170554424 cites W3009561768 @default.
- W3170554424 cites W3034445277 @default.
- W3170554424 cites W3034781633 @default.
- W3170554424 cites W3035524453 @default.
- W3170554424 cites W3095121901 @default.
- W3170554424 cites W3099206234 @default.
- W3170554424 cites W3100345210 @default.
- W3170554424 cites W3100859887 @default.
- W3170554424 cites W3101821705 @default.
- W3170554424 cites W3102363610 @default.
- W3170554424 cites W3106743555 @default.
- W3170554424 cites W3118608800 @default.
- W3170554424 cites W3122325173 @default.
- W3170554424 cites W343636949 @default.
- W3170554424 doi "https://doi.org/10.48550/arxiv.2106.03719" @default.
- W3170554424 hasPublicationYear "2021" @default.
- W3170554424 type Work @default.
- W3170554424 sameAs 3170554424 @default.
- W3170554424 citedByCount "0" @default.
- W3170554424 crossrefType "posted-content" @default.
- W3170554424 hasAuthorship W3170554424A5009305504 @default.
- W3170554424 hasAuthorship W3170554424A5058383656 @default.
- W3170554424 hasAuthorship W3170554424A5063179713 @default.
- W3170554424 hasAuthorship W3170554424A5064935846 @default.
- W3170554424 hasAuthorship W3170554424A5090586839 @default.
- W3170554424 hasBestOaLocation W31705544241 @default.
- W3170554424 hasConcept C111919701 @default.
- W3170554424 hasConcept C115961682 @default.
- W3170554424 hasConcept C118505674 @default.
- W3170554424 hasConcept C119857082 @default.
- W3170554424 hasConcept C136389625 @default.
- W3170554424 hasConcept C153180895 @default.
- W3170554424 hasConcept C154945302 @default.
- W3170554424 hasConcept C204321447 @default.
- W3170554424 hasConcept C41008148 @default.
- W3170554424 hasConcept C41608201 @default.
- W3170554424 hasConcept C50644808 @default.
- W3170554424 hasConcept C98045186 @default.
- W3170554424 hasConceptScore W3170554424C111919701 @default.
- W3170554424 hasConceptScore W3170554424C115961682 @default.
- W3170554424 hasConceptScore W3170554424C118505674 @default.
- W3170554424 hasConceptScore W3170554424C119857082 @default.
- W3170554424 hasConceptScore W3170554424C136389625 @default.
- W3170554424 hasConceptScore W3170554424C153180895 @default.
- W3170554424 hasConceptScore W3170554424C154945302 @default.
- W3170554424 hasConceptScore W3170554424C204321447 @default.
- W3170554424 hasConceptScore W3170554424C41008148 @default.
- W3170554424 hasConceptScore W3170554424C41608201 @default.
- W3170554424 hasConceptScore W3170554424C50644808 @default.
- W3170554424 hasConceptScore W3170554424C98045186 @default.
- W3170554424 hasLocation W31705544241 @default.
- W3170554424 hasOpenAccess W3170554424 @default.
- W3170554424 hasPrimaryLocation W31705544241 @default.
- W3170554424 hasRelatedWork W2794527914 @default.
- W3170554424 hasRelatedWork W3095538999 @default.