Matches in SemOpenAlex for { <https://semopenalex.org/work/W4308022412> ?p ?o ?g. }
Showing items 1 to 72 of
72
with 100 items per page.
- W4308022412 abstract "Data augmentation is a widely used technique and an essential ingredient in the recent advance in self-supervised representation learning. By preserving the similarity between augmented data, the resulting data representation can improve various downstream analyses and achieve state-of-art performance in many applications. To demystify the role of data augmentation, we develop a statistical framework on a low-dimension product manifold to theoretically understand why the unlabeled augmented data can lead to useful data representation. Under this framework, we propose a new representation learning method called augmentation invariant manifold learning and develop the corresponding loss function, which can work with a deep neural network to learn data representations. Compared with existing methods, the new data representation simultaneously exploits the manifold's geometric structure and invariant property of augmented data. Our theoretical investigation precisely characterizes how the data representation learned from augmented data can improve the $k$-nearest neighbor classifier in the downstream analysis, showing that a more complex data augmentation leads to more improvement in downstream analysis. Finally, numerical experiments on simulated and real datasets are presented to support the theoretical results in this paper." @default.
- W4308022412 created "2022-11-07" @default.
- W4308022412 creator A5078789660 @default.
- W4308022412 date "2022-11-01" @default.
- W4308022412 modified "2023-10-16" @default.
- W4308022412 title "Augmentation Invariant Manifold Learning" @default.
- W4308022412 doi "https://doi.org/10.48550/arxiv.2211.00460" @default.
- W4308022412 hasPublicationYear "2022" @default.
- W4308022412 type Work @default.
- W4308022412 citedByCount "0" @default.
- W4308022412 crossrefType "posted-content" @default.
- W4308022412 hasAuthorship W4308022412A5078789660 @default.
- W4308022412 hasBestOaLocation W43080224121 @default.
- W4308022412 hasConcept C111030470 @default.
- W4308022412 hasConcept C116409475 @default.
- W4308022412 hasConcept C119857082 @default.
- W4308022412 hasConcept C127413603 @default.
- W4308022412 hasConcept C151876577 @default.
- W4308022412 hasConcept C153180895 @default.
- W4308022412 hasConcept C154945302 @default.
- W4308022412 hasConcept C17744445 @default.
- W4308022412 hasConcept C190470478 @default.
- W4308022412 hasConcept C199539241 @default.
- W4308022412 hasConcept C2776359362 @default.
- W4308022412 hasConcept C30732413 @default.
- W4308022412 hasConcept C33923547 @default.
- W4308022412 hasConcept C37914503 @default.
- W4308022412 hasConcept C41008148 @default.
- W4308022412 hasConcept C529865628 @default.
- W4308022412 hasConcept C59404180 @default.
- W4308022412 hasConcept C70518039 @default.
- W4308022412 hasConcept C78519656 @default.
- W4308022412 hasConcept C94625758 @default.
- W4308022412 hasConcept C95623464 @default.
- W4308022412 hasConceptScore W4308022412C111030470 @default.
- W4308022412 hasConceptScore W4308022412C116409475 @default.
- W4308022412 hasConceptScore W4308022412C119857082 @default.
- W4308022412 hasConceptScore W4308022412C127413603 @default.
- W4308022412 hasConceptScore W4308022412C151876577 @default.
- W4308022412 hasConceptScore W4308022412C153180895 @default.
- W4308022412 hasConceptScore W4308022412C154945302 @default.
- W4308022412 hasConceptScore W4308022412C17744445 @default.
- W4308022412 hasConceptScore W4308022412C190470478 @default.
- W4308022412 hasConceptScore W4308022412C199539241 @default.
- W4308022412 hasConceptScore W4308022412C2776359362 @default.
- W4308022412 hasConceptScore W4308022412C30732413 @default.
- W4308022412 hasConceptScore W4308022412C33923547 @default.
- W4308022412 hasConceptScore W4308022412C37914503 @default.
- W4308022412 hasConceptScore W4308022412C41008148 @default.
- W4308022412 hasConceptScore W4308022412C529865628 @default.
- W4308022412 hasConceptScore W4308022412C59404180 @default.
- W4308022412 hasConceptScore W4308022412C70518039 @default.
- W4308022412 hasConceptScore W4308022412C78519656 @default.
- W4308022412 hasConceptScore W4308022412C94625758 @default.
- W4308022412 hasConceptScore W4308022412C95623464 @default.
- W4308022412 hasLocation W43080224121 @default.
- W4308022412 hasLocation W43080224122 @default.
- W4308022412 hasOpenAccess W4308022412 @default.
- W4308022412 hasPrimaryLocation W43080224121 @default.
- W4308022412 hasRelatedWork W1576868969 @default.
- W4308022412 hasRelatedWork W2163922914 @default.
- W4308022412 hasRelatedWork W2526784818 @default.
- W4308022412 hasRelatedWork W2540547653 @default.
- W4308022412 hasRelatedWork W2811036446 @default.
- W4308022412 hasRelatedWork W2905846897 @default.
- W4308022412 hasRelatedWork W3175304158 @default.
- W4308022412 hasRelatedWork W4300011764 @default.
- W4308022412 hasRelatedWork W4312396261 @default.
- W4308022412 hasRelatedWork W564581980 @default.
- W4308022412 isParatext "false" @default.
- W4308022412 isRetracted "false" @default.
- W4308022412 workType "article" @default.