Logotipo do repositório
 

Publicação:
Efficient Rank-Based Diffusion Process with Assured Convergence

dc.contributor.authorGuimaraes Pedronette, Daniel Carlos [UNESP]
dc.contributor.authorPascotti Valem, Lucas [UNESP]
dc.contributor.authorLatecki, Longin Jan
dc.contributor.institutionUniversidade Estadual Paulista (Unesp)
dc.contributor.institutionTemple Univ
dc.date.accessioned2021-06-25T12:41:32Z
dc.date.available2021-06-25T12:41:32Z
dc.date.issued2021-03-01
dc.description.abstractVisual features and representation learning strategies experienced huge advances in the previous decade, mainly supported by deep learning approaches. However, retrieval tasks are still performed mainly based on traditional pairwise dissimilarity measures, while the learned representations lie on high dimensional manifolds. With the aim of going beyond pairwise analysis, post-processing methods have been proposed to replace pairwise measures by globally defined measures, capable of analyzing collections in terms of the underlying data manifold. The most representative approaches are diffusion and ranked-based methods. While the diffusion approaches can be computationally expensive, the rank-based methods lack theoretical background. In this paper, we propose an efficient Rank-based Diffusion Process which combines both approaches and avoids the drawbacks of each one. The obtained method is capable of efficiently approximating a diffusion process by exploiting rank-based information, while assuring its convergence. The algorithm exhibits very low asymptotic complexity and can be computed regionally, being suitable to outside of dataset queries. An experimental evaluation conducted for image retrieval and person re-ID tasks on diverse datasets demonstrates the effectiveness of the proposed approach with results comparable to the state-of-the-art.en
dc.description.affiliationSao Paulo State Univ UNESP, Dept Stat Appl Math & Comp DEMAC, BR-13506900 Rio Claro, Brazil
dc.description.affiliationTemple Univ, Dept Comp & Informat Sci, Philadelphia, PA 19122 USA
dc.description.affiliationUnespSao Paulo State Univ UNESP, Dept Stat Appl Math & Comp DEMAC, BR-13506900 Rio Claro, Brazil
dc.description.sponsorshipFundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
dc.description.sponsorshipConselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
dc.description.sponsorshipMicrosoft Research
dc.description.sponsorshipNational Science Foundation
dc.description.sponsorshipFulbright Commission
dc.description.sponsorshipIdFAPESP: 2018/15597-6
dc.description.sponsorshipIdFAPESP: 2017/25908-6
dc.description.sponsorshipIdFAPESP: 2020/11366-0
dc.description.sponsorshipIdCNPq: 308194/2017-9
dc.description.sponsorshipIdCNPq: 309439/2020-5
dc.description.sponsorshipIdNational Science Foundation: IIS-1814745
dc.format.extent23
dc.identifierhttp://dx.doi.org/10.3390/jimaging7030049
dc.identifier.citationJournal Of Imaging. Basel: Mdpi, v. 7, n. 3, 23 p., 2021.
dc.identifier.doi10.3390/jimaging7030049
dc.identifier.urihttp://hdl.handle.net/11449/210161
dc.identifier.wosWOS:000633781900001
dc.language.isoeng
dc.publisherMdpi
dc.relation.ispartofJournal Of Imaging
dc.sourceWeb of Science
dc.subjectdiffusion
dc.subjectrank
dc.subjectimage retrieval
dc.subjectconvergence
dc.titleEfficient Rank-Based Diffusion Process with Assured Convergenceen
dc.typeArtigo
dcterms.rightsHolderMdpi
dspace.entity.typePublication
unesp.campusUniversidade Estadual Paulista (UNESP), Instituto de Geociências e Ciências Exatas, Rio Claropt
unesp.departmentEstatística, Matemática Aplicada e Computação - IGCEpt

Arquivos