Matches in SemOpenAlex for { <https://semopenalex.org/work/W3207047134> ?p ?o ?g. }
- W3207047134 abstract "The pretrain-finetune paradigm has shown outstanding performance on many applications of deep learning, where a model is pre-trained on a upstream large dataset (e.g. ImageNet), and is then fine-tuned to different downstream tasks. Though for most cases, the pre-training stage is conducted based on supervised methods, recent works on self-supervised pre-training have shown powerful transferability and even outperform supervised pre-training on multiple downstream tasks. It thus remains an open question how to better generalize supervised pre-training model to downstream tasks. In this paper, we argue that the worse transferability of existing supervised pre-training methods arise from the negligence of valuable intra-class semantic difference. This is because these methods tend to push images from the same class close to each other despite of the large diversity in their visual contents, a problem to which referred as overfit of upstream tasks. To alleviate this problem, we propose a new supervised pre-training method based on Leave-One-Out K-Nearest-Neighbor, or LOOK for short. It relieves the problem of overfitting upstream tasks by only requiring each image to share its class label with most of its k nearest neighbors, thus allowing each class to exhibit a multi-mode distribution and consequentially preserving part of intra-class difference for better transferring to downstream tasks. We developed efficient implementation of the proposed method that scales well to large datasets. Experimental studies on multiple downstream tasks show that LOOK outperforms other state-of-the-art methods for supervised and self-supervised pre-training." @default.
- W3207047134 created "2021-10-25" @default.
- W3207047134 creator A5001956754 @default.
- W3207047134 creator A5020379648 @default.
- W3207047134 creator A5034814310 @default.
- W3207047134 creator A5060450036 @default.
- W3207047134 creator A5069394608 @default.
- W3207047134 date "2021-10-12" @default.
- W3207047134 modified "2023-10-16" @default.
- W3207047134 title "Rethinking Supervised Pre-training for Better Downstream Transferring" @default.
- W3207047134 cites W1686810756 @default.
- W3207047134 cites W1846799578 @default.
- W3207047134 cites W2047643928 @default.
- W3207047134 cites W2099001231 @default.
- W3207047134 cites W2108598243 @default.
- W3207047134 cites W2138011018 @default.
- W3207047134 cites W2149933564 @default.
- W3207047134 cites W2163605009 @default.
- W3207047134 cites W2187089797 @default.
- W3207047134 cites W2194321275 @default.
- W3207047134 cites W2194775991 @default.
- W3207047134 cites W2321533354 @default.
- W3207047134 cites W2533598788 @default.
- W3207047134 cites W2536397232 @default.
- W3207047134 cites W2618799552 @default.
- W3207047134 cites W2883725317 @default.
- W3207047134 cites W2887280559 @default.
- W3207047134 cites W2932083555 @default.
- W3207047134 cites W2948433173 @default.
- W3207047134 cites W2962742544 @default.
- W3207047134 cites W2962843773 @default.
- W3207047134 cites W2963150697 @default.
- W3207047134 cites W2963259451 @default.
- W3207047134 cites W2963341956 @default.
- W3207047134 cites W2963524571 @default.
- W3207047134 cites W2963826681 @default.
- W3207047134 cites W2964186069 @default.
- W3207047134 cites W2964194231 @default.
- W3207047134 cites W2965373594 @default.
- W3207047134 cites W2971339032 @default.
- W3207047134 cites W2995181141 @default.
- W3207047134 cites W3008769831 @default.
- W3207047134 cites W3009561768 @default.
- W3207047134 cites W3034748927 @default.
- W3207047134 cites W3034978746 @default.
- W3207047134 cites W3035524453 @default.
- W3207047134 cites W3094502228 @default.
- W3207047134 cites W3097217077 @default.
- W3207047134 cites W3098903812 @default.
- W3207047134 cites W3100345210 @default.
- W3207047134 cites W3103516055 @default.
- W3207047134 cites W3122325173 @default.
- W3207047134 cites W3134652006 @default.
- W3207047134 cites W3137952714 @default.
- W3207047134 cites W3170863103 @default.
- W3207047134 cites W3171007011 @default.
- W3207047134 cites W639708223 @default.
- W3207047134 doi "https://doi.org/10.48550/arxiv.2110.06014" @default.
- W3207047134 hasPublicationYear "2021" @default.
- W3207047134 type Work @default.
- W3207047134 sameAs 3207047134 @default.
- W3207047134 citedByCount "0" @default.
- W3207047134 crossrefType "posted-content" @default.
- W3207047134 hasAuthorship W3207047134A5001956754 @default.
- W3207047134 hasAuthorship W3207047134A5020379648 @default.
- W3207047134 hasAuthorship W3207047134A5034814310 @default.
- W3207047134 hasAuthorship W3207047134A5060450036 @default.
- W3207047134 hasAuthorship W3207047134A5069394608 @default.
- W3207047134 hasBestOaLocation W32070471341 @default.
- W3207047134 hasConcept C119857082 @default.
- W3207047134 hasConcept C121332964 @default.
- W3207047134 hasConcept C127413603 @default.
- W3207047134 hasConcept C136389625 @default.
- W3207047134 hasConcept C140331021 @default.
- W3207047134 hasConcept C153180895 @default.
- W3207047134 hasConcept C153294291 @default.
- W3207047134 hasConcept C154945302 @default.
- W3207047134 hasConcept C191172861 @default.
- W3207047134 hasConcept C21547014 @default.
- W3207047134 hasConcept C22019652 @default.
- W3207047134 hasConcept C2776207758 @default.
- W3207047134 hasConcept C2777211547 @default.
- W3207047134 hasConcept C2777212361 @default.
- W3207047134 hasConcept C31258907 @default.
- W3207047134 hasConcept C41008148 @default.
- W3207047134 hasConcept C50644808 @default.
- W3207047134 hasConcept C61272859 @default.
- W3207047134 hasConceptScore W3207047134C119857082 @default.
- W3207047134 hasConceptScore W3207047134C121332964 @default.
- W3207047134 hasConceptScore W3207047134C127413603 @default.
- W3207047134 hasConceptScore W3207047134C136389625 @default.
- W3207047134 hasConceptScore W3207047134C140331021 @default.
- W3207047134 hasConceptScore W3207047134C153180895 @default.
- W3207047134 hasConceptScore W3207047134C153294291 @default.
- W3207047134 hasConceptScore W3207047134C154945302 @default.
- W3207047134 hasConceptScore W3207047134C191172861 @default.
- W3207047134 hasConceptScore W3207047134C21547014 @default.
- W3207047134 hasConceptScore W3207047134C22019652 @default.
- W3207047134 hasConceptScore W3207047134C2776207758 @default.
- W3207047134 hasConceptScore W3207047134C2777211547 @default.