DC Field | Value | Language |
---|---|---|
dc.contributor.author | KIM, BYUNG HYUNG | ko |
dc.contributor.author | Suh,Y | ko |
dc.contributor.author | Lee,H | ko |
dc.contributor.author | Jo, Sung-Ho | ko |
dc.date.accessioned | 2021-02-17T05:10:14Z | - |
dc.date.available | 2021-02-17T05:10:14Z | - |
dc.date.created | 2020-10-21 | - |
dc.date.created | 2020-10-21 | - |
dc.date.created | 2020-10-21 | - |
dc.date.issued | 2021-01 | - |
dc.identifier.citation | 25th International Conference on Pattern Recognition (ICPR) , pp.4348 - 4355 | - |
dc.identifier.issn | 1051-4651 | - |
dc.identifier.uri | http://hdl.handle.net/10203/280795 | - |
dc.description.abstract | We propose a rank-based metric learning method by leveraging a concept of the Riemannian Potato for better separating non-linear data. By exploring the geometric properties of Riemannian manifolds, the proposed loss function optimizes the measure of dispersion using the distribution of Riemannian distances between a reference sample and neighbors and builds a ranked list according to the similarities. We show the proposed function can learn a hypersphere for each class, preserving the similarity structure inside it on Riemannian manifold. As a result, compared with Euclidean distance-based metric, our method can further jointly reduce the intra-class distances and enlarge the inter-class distances for learned features, consistently outperforming state-of-the-art methods on three widely used nonlinear datasets. | - |
dc.language | English | - |
dc.publisher | International Association of Pattern Recognition | - |
dc.title | Nonlinear Ranking Loss on Riemannian Potato Embedding | - |
dc.type | Conference | - |
dc.identifier.wosid | 000678409204063 | - |
dc.identifier.scopusid | 2-s2.0-85104879427 | - |
dc.type.rims | CONF | - |
dc.citation.beginningpage | 4348 | - |
dc.citation.endingpage | 4355 | - |
dc.citation.publicationname | 25th International Conference on Pattern Recognition (ICPR) | - |
dc.identifier.conferencecountry | IT | - |
dc.identifier.conferencelocation | Virtual | - |
dc.identifier.doi | 10.1109/ICPR48806.2021.9412664 | - |
dc.contributor.localauthor | KIM, BYUNG HYUNG | - |
dc.contributor.localauthor | Jo, Sung-Ho | - |
dc.contributor.nonIdAuthor | Suh,Y | - |
dc.contributor.nonIdAuthor | Lee,H | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.