Matches in SemOpenAlex for { <https://semopenalex.org/work/W2962762068> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W2962762068 abstract "In this paper, we study the problem of learning image classification models with label noise. Existing approaches depending on human supervision are generally not scalable as manually identifying correct or incorrect labels is time-consuming, whereas approaches not relying on human supervision are scalable but less effective. To reduce the amount of human supervision for label noise cleaning, we introduce CleanNet, a joint neural embedding network, which only requires a fraction of the classes being manually verified to provide the knowledge of label noise that can be transferred to other classes. We further integrate CleanNet and conventional convolutional neural network classifier into one framework for image classification learning. We demonstrate the effectiveness of the proposed algorithm on both of the label noise detection task and the image classification on noisy data task on several large-scale datasets. Experimental results show that CleanNet can reduce label noise detection error rate on held-out classes where no human supervision available by 41.5% compared to current weakly supervised methods. It also achieves 47% of the performance gain of verifying all images with only 3.2% images verified on an image classification task. Source code and dataset will be available at kuanghuei.github.io/CleanNetProject." @default.
- W2962762068 created "2019-07-30" @default.
- W2962762068 creator A5006832020 @default.
- W2962762068 creator A5016594236 @default.
- W2962762068 creator A5048159038 @default.
- W2962762068 creator A5065781568 @default.
- W2962762068 date "2018-06-01" @default.
- W2962762068 modified "2023-10-17" @default.
- W2962762068 title "CleanNet: Transfer Learning for Scalable Image Classifier Training with Label Noise" @default.
- W2962762068 cites W1994550352 @default.
- W2962762068 cites W2027444427 @default.
- W2962762068 cites W2081580037 @default.
- W2962762068 cites W2100495367 @default.
- W2962762068 cites W2108598243 @default.
- W2962762068 cites W2122146326 @default.
- W2962762068 cites W2124219775 @default.
- W2962762068 cites W2132870739 @default.
- W2962762068 cites W2147483361 @default.
- W2962762068 cites W2167460663 @default.
- W2962762068 cites W2194775991 @default.
- W2962762068 cites W2204904589 @default.
- W2962762068 cites W2470673105 @default.
- W2962762068 cites W2557912764 @default.
- W2962762068 cites W2577784528 @default.
- W2962762068 cites W2592335154 @default.
- W2962762068 cites W2593768305 @default.
- W2962762068 cites W2603705233 @default.
- W2962762068 cites W2737041163 @default.
- W2962762068 cites W2964292098 @default.
- W2962762068 doi "https://doi.org/10.1109/cvpr.2018.00571" @default.
- W2962762068 hasPublicationYear "2018" @default.
- W2962762068 type Work @default.
- W2962762068 sameAs 2962762068 @default.
- W2962762068 citedByCount "293" @default.
- W2962762068 countsByYear W29627620682017 @default.
- W2962762068 countsByYear W29627620682018 @default.
- W2962762068 countsByYear W29627620682019 @default.
- W2962762068 countsByYear W29627620682020 @default.
- W2962762068 countsByYear W29627620682021 @default.
- W2962762068 countsByYear W29627620682022 @default.
- W2962762068 countsByYear W29627620682023 @default.
- W2962762068 crossrefType "proceedings-article" @default.
- W2962762068 hasAuthorship W2962762068A5006832020 @default.
- W2962762068 hasAuthorship W2962762068A5016594236 @default.
- W2962762068 hasAuthorship W2962762068A5048159038 @default.
- W2962762068 hasAuthorship W2962762068A5065781568 @default.
- W2962762068 hasBestOaLocation W29627620682 @default.
- W2962762068 hasConcept C115961682 @default.
- W2962762068 hasConcept C119857082 @default.
- W2962762068 hasConcept C150899416 @default.
- W2962762068 hasConcept C153180895 @default.
- W2962762068 hasConcept C154945302 @default.
- W2962762068 hasConcept C163294075 @default.
- W2962762068 hasConcept C29265498 @default.
- W2962762068 hasConcept C41008148 @default.
- W2962762068 hasConcept C41608201 @default.
- W2962762068 hasConcept C48044578 @default.
- W2962762068 hasConcept C50644808 @default.
- W2962762068 hasConcept C75294576 @default.
- W2962762068 hasConcept C77088390 @default.
- W2962762068 hasConcept C81363708 @default.
- W2962762068 hasConcept C95623464 @default.
- W2962762068 hasConcept C99498987 @default.
- W2962762068 hasConceptScore W2962762068C115961682 @default.
- W2962762068 hasConceptScore W2962762068C119857082 @default.
- W2962762068 hasConceptScore W2962762068C150899416 @default.
- W2962762068 hasConceptScore W2962762068C153180895 @default.
- W2962762068 hasConceptScore W2962762068C154945302 @default.
- W2962762068 hasConceptScore W2962762068C163294075 @default.
- W2962762068 hasConceptScore W2962762068C29265498 @default.
- W2962762068 hasConceptScore W2962762068C41008148 @default.
- W2962762068 hasConceptScore W2962762068C41608201 @default.
- W2962762068 hasConceptScore W2962762068C48044578 @default.
- W2962762068 hasConceptScore W2962762068C50644808 @default.
- W2962762068 hasConceptScore W2962762068C75294576 @default.
- W2962762068 hasConceptScore W2962762068C77088390 @default.
- W2962762068 hasConceptScore W2962762068C81363708 @default.
- W2962762068 hasConceptScore W2962762068C95623464 @default.
- W2962762068 hasConceptScore W2962762068C99498987 @default.
- W2962762068 hasLocation W29627620681 @default.
- W2962762068 hasLocation W29627620682 @default.
- W2962762068 hasOpenAccess W2962762068 @default.
- W2962762068 hasPrimaryLocation W29627620681 @default.
- W2962762068 hasRelatedWork W2760944304 @default.
- W2962762068 hasRelatedWork W2909857627 @default.
- W2962762068 hasRelatedWork W2972069047 @default.
- W2962762068 hasRelatedWork W2995914718 @default.
- W2962762068 hasRelatedWork W3012393889 @default.
- W2962762068 hasRelatedWork W3014041368 @default.
- W2962762068 hasRelatedWork W3018421652 @default.
- W2962762068 hasRelatedWork W3021430260 @default.
- W2962762068 hasRelatedWork W3189091156 @default.
- W2962762068 hasRelatedWork W564581980 @default.
- W2962762068 isParatext "false" @default.
- W2962762068 isRetracted "false" @default.
- W2962762068 magId "2962762068" @default.
- W2962762068 workType "article" @default.