Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386065630> ?p ?o ?g. }
Showing items 1 to 91 of
91
with 100 items per page.
- W4386065630 abstract "The success of deep learning is largely attributed to the training over clean data. However, data is often coupled with noisy labels in practice. Learning with noisy labels is challenging because the performance of the deep neural networks (DNN) drastically degenerates, due to confirmation bias caused by the network memorization over noisy labels. To alleviate that, a recent prominent direction is on sample selection, which retrieves clean data samples from noisy samples, so as to enhance the model's robustness and tolerance to noisy labels. In this paper, we re-vamp the sample selection from the perspective of optimal transport theory and propose a novel method, called the OT-Filter. The OT-Filter provides geometrically meaningful distances and preserves distribution patterns to measure the data discrepancy, thus alleviating the confirmation bias. Extensive experiments on benchmarks, such as Clothing1M and ANIMAL-10N, show that the performance of the OT-Filter outperforms its counterparts. Meanwhile, results on benchmarks with synthetic labels, such as CIFAR-10/100, show the superiority of the OT-Filter in handling data labels of high noise." @default.
- W4386065630 created "2023-08-23" @default.
- W4386065630 creator A5037366245 @default.
- W4386065630 creator A5059358267 @default.
- W4386065630 creator A5089069072 @default.
- W4386065630 date "2023-06-01" @default.
- W4386065630 modified "2023-09-26" @default.
- W4386065630 title "OT-Filter: An Optimal Transport Filter for Learning with Noisy Labels" @default.
- W4386065630 cites W1968333723 @default.
- W4386065630 cites W2036996178 @default.
- W4386065630 cites W2049633694 @default.
- W4386065630 cites W2079639858 @default.
- W4386065630 cites W2108598243 @default.
- W4386065630 cites W2194775991 @default.
- W4386065630 cites W2298250094 @default.
- W4386065630 cites W2585226541 @default.
- W4386065630 cites W2895694115 @default.
- W4386065630 cites W2964292098 @default.
- W4386065630 cites W3096461453 @default.
- W4386065630 cites W3100137005 @default.
- W4386065630 cites W3171635822 @default.
- W4386065630 cites W3175103763 @default.
- W4386065630 cites W3175831284 @default.
- W4386065630 cites W3181871665 @default.
- W4386065630 cites W3186514747 @default.
- W4386065630 cites W3205655893 @default.
- W4386065630 cites W4206471589 @default.
- W4386065630 cites W4221161368 @default.
- W4386065630 cites W4233762729 @default.
- W4386065630 cites W4250589301 @default.
- W4386065630 cites W4255839052 @default.
- W4386065630 cites W4312305885 @default.
- W4386065630 cites W4312440249 @default.
- W4386065630 cites W4312766345 @default.
- W4386065630 doi "https://doi.org/10.1109/cvpr52729.2023.01551" @default.
- W4386065630 hasPublicationYear "2023" @default.
- W4386065630 type Work @default.
- W4386065630 citedByCount "0" @default.
- W4386065630 crossrefType "proceedings-article" @default.
- W4386065630 hasAuthorship W4386065630A5037366245 @default.
- W4386065630 hasAuthorship W4386065630A5059358267 @default.
- W4386065630 hasAuthorship W4386065630A5089069072 @default.
- W4386065630 hasConcept C104317684 @default.
- W4386065630 hasConcept C106131492 @default.
- W4386065630 hasConcept C115961682 @default.
- W4386065630 hasConcept C119857082 @default.
- W4386065630 hasConcept C145420912 @default.
- W4386065630 hasConcept C153180895 @default.
- W4386065630 hasConcept C154945302 @default.
- W4386065630 hasConcept C185592680 @default.
- W4386065630 hasConcept C2781170535 @default.
- W4386065630 hasConcept C30038468 @default.
- W4386065630 hasConcept C31972630 @default.
- W4386065630 hasConcept C33923547 @default.
- W4386065630 hasConcept C41008148 @default.
- W4386065630 hasConcept C55493867 @default.
- W4386065630 hasConcept C63479239 @default.
- W4386065630 hasConcept C99498987 @default.
- W4386065630 hasConceptScore W4386065630C104317684 @default.
- W4386065630 hasConceptScore W4386065630C106131492 @default.
- W4386065630 hasConceptScore W4386065630C115961682 @default.
- W4386065630 hasConceptScore W4386065630C119857082 @default.
- W4386065630 hasConceptScore W4386065630C145420912 @default.
- W4386065630 hasConceptScore W4386065630C153180895 @default.
- W4386065630 hasConceptScore W4386065630C154945302 @default.
- W4386065630 hasConceptScore W4386065630C185592680 @default.
- W4386065630 hasConceptScore W4386065630C2781170535 @default.
- W4386065630 hasConceptScore W4386065630C30038468 @default.
- W4386065630 hasConceptScore W4386065630C31972630 @default.
- W4386065630 hasConceptScore W4386065630C33923547 @default.
- W4386065630 hasConceptScore W4386065630C41008148 @default.
- W4386065630 hasConceptScore W4386065630C55493867 @default.
- W4386065630 hasConceptScore W4386065630C63479239 @default.
- W4386065630 hasConceptScore W4386065630C99498987 @default.
- W4386065630 hasFunder F4320321001 @default.
- W4386065630 hasLocation W43860656301 @default.
- W4386065630 hasOpenAccess W4386065630 @default.
- W4386065630 hasPrimaryLocation W43860656301 @default.
- W4386065630 hasRelatedWork W2899084033 @default.
- W4386065630 hasRelatedWork W2961085424 @default.
- W4386065630 hasRelatedWork W3046775127 @default.
- W4386065630 hasRelatedWork W3170094116 @default.
- W4386065630 hasRelatedWork W4205958290 @default.
- W4386065630 hasRelatedWork W4285260836 @default.
- W4386065630 hasRelatedWork W4286629047 @default.
- W4386065630 hasRelatedWork W4306321456 @default.
- W4386065630 hasRelatedWork W4306674287 @default.
- W4386065630 hasRelatedWork W4224009465 @default.
- W4386065630 isParatext "false" @default.
- W4386065630 isRetracted "false" @default.
- W4386065630 workType "article" @default.