Matches in SemOpenAlex for { <https://semopenalex.org/work/W2979302305> ?p ?o ?g. }
Showing items 1 to 89 of
89
with 100 items per page.
- W2979302305 abstract "The predictive performance of supervised learning algorithms depends on the quality of labels. In a typical label collection process, multiple annotators provide subjective noisy estimates of the ``truth under the influence of their varying skill-levels and biases. Blindly treating these noisy labels as the ground truth limits the accuracy of learning algorithms in the presence of strong disagreement. This problem is critical for applications in domains such as medical imaging where both the annotation cost and inter-observer variability are high. In this work, we present a method for simultaneously learning the individual annotator model and the underlying true label distribution, using only noisy observations. Each annotator is modeled by a confusion matrix that is jointly estimated along with the classifier predictions. We propose to add a regularization term to the loss function that encourages convergence to the true annotator confusion matrix. We provide a theoretical argument as to how the regularization is essential to our approach both for the case of single annotator and multiple annotators. Despite the simplicity of the idea, experiments on image classification tasks with both simulated and real labels show that our method either outperforms or performs on par with the state-of-the-art methods and is capable of estimating the skills of annotators even with a single label available per image." @default.
- W2979302305 created "2019-10-18" @default.
- W2979302305 creator A5033449704 @default.
- W2979302305 creator A5037303942 @default.
- W2979302305 creator A5064157732 @default.
- W2979302305 creator A5081528888 @default.
- W2979302305 creator A5084552262 @default.
- W2979302305 date "2019-06-01" @default.
- W2979302305 modified "2023-10-10" @default.
- W2979302305 title "Learning From Noisy Labels by Regularized Estimation of Annotator Confusion" @default.
- W2979302305 cites W1980933880 @default.
- W2979302305 cites W2010135967 @default.
- W2979302305 cites W2029327121 @default.
- W2979302305 cites W2097036415 @default.
- W2979302305 cites W2135761719 @default.
- W2979302305 cites W2148347694 @default.
- W2979302305 cites W2150612552 @default.
- W2979302305 cites W2164124780 @default.
- W2979302305 cites W2167460663 @default.
- W2979302305 cites W2201054263 @default.
- W2979302305 cites W2577784528 @default.
- W2979302305 cites W2592929672 @default.
- W2979302305 cites W2745544219 @default.
- W2979302305 cites W2798896881 @default.
- W2979302305 cites W2964274690 @default.
- W2979302305 cites W2964292098 @default.
- W2979302305 cites W72116547 @default.
- W2979302305 cites W9014458 @default.
- W2979302305 doi "https://doi.org/10.1109/cvpr.2019.01150" @default.
- W2979302305 hasPublicationYear "2019" @default.
- W2979302305 type Work @default.
- W2979302305 sameAs 2979302305 @default.
- W2979302305 citedByCount "113" @default.
- W2979302305 countsByYear W29793023052019 @default.
- W2979302305 countsByYear W29793023052020 @default.
- W2979302305 countsByYear W29793023052021 @default.
- W2979302305 countsByYear W29793023052022 @default.
- W2979302305 countsByYear W29793023052023 @default.
- W2979302305 crossrefType "proceedings-article" @default.
- W2979302305 hasAuthorship W2979302305A5033449704 @default.
- W2979302305 hasAuthorship W2979302305A5037303942 @default.
- W2979302305 hasAuthorship W2979302305A5064157732 @default.
- W2979302305 hasAuthorship W2979302305A5081528888 @default.
- W2979302305 hasAuthorship W2979302305A5084552262 @default.
- W2979302305 hasBestOaLocation W29793023052 @default.
- W2979302305 hasConcept C11171543 @default.
- W2979302305 hasConcept C119857082 @default.
- W2979302305 hasConcept C138602881 @default.
- W2979302305 hasConcept C146849305 @default.
- W2979302305 hasConcept C153180895 @default.
- W2979302305 hasConcept C154945302 @default.
- W2979302305 hasConcept C15744967 @default.
- W2979302305 hasConcept C2776135515 @default.
- W2979302305 hasConcept C2776321320 @default.
- W2979302305 hasConcept C2781140086 @default.
- W2979302305 hasConcept C41008148 @default.
- W2979302305 hasConcept C95623464 @default.
- W2979302305 hasConceptScore W2979302305C11171543 @default.
- W2979302305 hasConceptScore W2979302305C119857082 @default.
- W2979302305 hasConceptScore W2979302305C138602881 @default.
- W2979302305 hasConceptScore W2979302305C146849305 @default.
- W2979302305 hasConceptScore W2979302305C153180895 @default.
- W2979302305 hasConceptScore W2979302305C154945302 @default.
- W2979302305 hasConceptScore W2979302305C15744967 @default.
- W2979302305 hasConceptScore W2979302305C2776135515 @default.
- W2979302305 hasConceptScore W2979302305C2776321320 @default.
- W2979302305 hasConceptScore W2979302305C2781140086 @default.
- W2979302305 hasConceptScore W2979302305C41008148 @default.
- W2979302305 hasConceptScore W2979302305C95623464 @default.
- W2979302305 hasLocation W29793023051 @default.
- W2979302305 hasLocation W29793023052 @default.
- W2979302305 hasLocation W29793023053 @default.
- W2979302305 hasLocation W29793023054 @default.
- W2979302305 hasOpenAccess W2979302305 @default.
- W2979302305 hasPrimaryLocation W29793023051 @default.
- W2979302305 hasRelatedWork W191014748 @default.
- W2979302305 hasRelatedWork W2052067190 @default.
- W2979302305 hasRelatedWork W2069375627 @default.
- W2979302305 hasRelatedWork W2356966757 @default.
- W2979302305 hasRelatedWork W2570625548 @default.
- W2979302305 hasRelatedWork W2899196211 @default.
- W2979302305 hasRelatedWork W2961085424 @default.
- W2979302305 hasRelatedWork W3011378683 @default.
- W2979302305 hasRelatedWork W3160713586 @default.
- W2979302305 hasRelatedWork W4361795583 @default.
- W2979302305 isParatext "false" @default.
- W2979302305 isRetracted "false" @default.
- W2979302305 magId "2979302305" @default.
- W2979302305 workType "article" @default.