Matches in SemOpenAlex for { <https://semopenalex.org/work/W4387068052> ?p ?o ?g. }
Showing items 1 to 68 of
68
with 100 items per page.
- W4387068052 endingPage "18" @default.
- W4387068052 startingPage "1" @default.
- W4387068052 abstract "Positive-Unlabeled (PU) data arise frequently in a wide range of fields such as medical diagnosis, anomaly analysis and personalized advertising. The absence of any known negative labels makes it very challenging to learn binary classifiers from such data. Many state-of-the-art methods reformulate the original classification risk with individual risks over positive and unlabeled data, and explicitly minimize the risk of classifying unlabeled data as negative. This, however, usually leads to classifiers with a bias toward negative predictions, <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>i.e.</i> , they tend to recognize most unlabeled data as negative. In this paper, we propose a label distribution alignment formulation for PU learning to alleviate this issue. Specifically, we align the distribution of predicted labels with the ground-truth, which is constant for a given class prior. In this way, the proportion of samples predicted as negative is explicitly controlled from a global perspective, and thus the bias toward negative predictions could be intrinsically eliminated. On top of this, we further introduce the idea of functional margins to enhance the model's discriminability, and derive a margin-based learning framework named <italic xmlns:mml=http://www.w3.org/1998/Math/MathML xmlns:xlink=http://www.w3.org/1999/xlink>Positive-Unlabeled learning with Label Distribution Alignment (PULDA)</i> . This framework is also combined with the class prior estimation process for practical scenarios, and theoretically supported by a generalization analysis. Moreover, a stochastic mini-batch optimization algorithm based on the exponential moving average strategy is tailored for this problem with a convergence guarantee. Finally, comprehensive empirical results demonstrate the effectiveness of the proposed method." @default.
- W4387068052 created "2023-09-27" @default.
- W4387068052 creator A5009753955 @default.
- W4387068052 creator A5018138384 @default.
- W4387068052 creator A5028597017 @default.
- W4387068052 creator A5053213796 @default.
- W4387068052 creator A5059594320 @default.
- W4387068052 creator A5068837264 @default.
- W4387068052 creator A5089092566 @default.
- W4387068052 date "2023-01-01" @default.
- W4387068052 modified "2023-10-16" @default.
- W4387068052 title "Positive-Unlabeled Learning With Label Distribution Alignment" @default.
- W4387068052 doi "https://doi.org/10.1109/tpami.2023.3319431" @default.
- W4387068052 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/37751347" @default.
- W4387068052 hasPublicationYear "2023" @default.
- W4387068052 type Work @default.
- W4387068052 citedByCount "0" @default.
- W4387068052 crossrefType "journal-article" @default.
- W4387068052 hasAuthorship W4387068052A5009753955 @default.
- W4387068052 hasAuthorship W4387068052A5018138384 @default.
- W4387068052 hasAuthorship W4387068052A5028597017 @default.
- W4387068052 hasAuthorship W4387068052A5053213796 @default.
- W4387068052 hasAuthorship W4387068052A5059594320 @default.
- W4387068052 hasAuthorship W4387068052A5068837264 @default.
- W4387068052 hasAuthorship W4387068052A5089092566 @default.
- W4387068052 hasConcept C119857082 @default.
- W4387068052 hasConcept C134306372 @default.
- W4387068052 hasConcept C153180895 @default.
- W4387068052 hasConcept C154945302 @default.
- W4387068052 hasConcept C159985019 @default.
- W4387068052 hasConcept C177148314 @default.
- W4387068052 hasConcept C192562407 @default.
- W4387068052 hasConcept C204323151 @default.
- W4387068052 hasConcept C2777212361 @default.
- W4387068052 hasConcept C33923547 @default.
- W4387068052 hasConcept C41008148 @default.
- W4387068052 hasConcept C774472 @default.
- W4387068052 hasConceptScore W4387068052C119857082 @default.
- W4387068052 hasConceptScore W4387068052C134306372 @default.
- W4387068052 hasConceptScore W4387068052C153180895 @default.
- W4387068052 hasConceptScore W4387068052C154945302 @default.
- W4387068052 hasConceptScore W4387068052C159985019 @default.
- W4387068052 hasConceptScore W4387068052C177148314 @default.
- W4387068052 hasConceptScore W4387068052C192562407 @default.
- W4387068052 hasConceptScore W4387068052C204323151 @default.
- W4387068052 hasConceptScore W4387068052C2777212361 @default.
- W4387068052 hasConceptScore W4387068052C33923547 @default.
- W4387068052 hasConceptScore W4387068052C41008148 @default.
- W4387068052 hasConceptScore W4387068052C774472 @default.
- W4387068052 hasLocation W43870680521 @default.
- W4387068052 hasLocation W43870680522 @default.
- W4387068052 hasOpenAccess W4387068052 @default.
- W4387068052 hasPrimaryLocation W43870680521 @default.
- W4387068052 hasRelatedWork W2603409575 @default.
- W4387068052 hasRelatedWork W2885094885 @default.
- W4387068052 hasRelatedWork W2961085424 @default.
- W4387068052 hasRelatedWork W2979236518 @default.
- W4387068052 hasRelatedWork W3046775127 @default.
- W4387068052 hasRelatedWork W4285260836 @default.
- W4387068052 hasRelatedWork W4286629047 @default.
- W4387068052 hasRelatedWork W4306321456 @default.
- W4387068052 hasRelatedWork W4306674287 @default.
- W4387068052 hasRelatedWork W4224009465 @default.
- W4387068052 isParatext "false" @default.
- W4387068052 isRetracted "false" @default.
- W4387068052 workType "article" @default.