Inductive Self-Supervised Dimensionality Reduction for Image Retrieval

Registro completo de metadados
MetadadosDescriçãoIdioma
Autor(es): dc.contributorUniversidade Estadual Paulista (UNESP)-
Autor(es): dc.creatorBiotto, Deryk Willyan-
Autor(es): dc.creatorJardim, Guilherme Henrique-
Autor(es): dc.creatorKawai, Vinicius Atsushi Sato-
Autor(es): dc.creatorRozin, Bionda-
Autor(es): dc.creatorSalvadeo, Denis Henrique Pinheiro-
Autor(es): dc.creatorPedronette, Daniel Carlos Guimarães-
Data de aceite: dc.date.accessioned2025-08-21T23:43:57Z-
Data de disponibilização: dc.date.available2025-08-21T23:43:57Z-
Data de envio: dc.date.issued2025-04-29-
Data de envio: dc.date.issued2024-12-31-
Fonte completa do material: dc.identifierhttp://dx.doi.org/10.5220/0013158600003912-
Fonte completa do material: dc.identifierhttps://hdl.handle.net/11449/306793-
Fonte: dc.identifier.urihttp://educapes.capes.gov.br/handle/11449/306793-
Descrição: dc.descriptionThe exponential growth of multimidia data creates a pressing need for approaches that are capable of efficiently handling Content-Based Image Retrieval (CBIR) in large and continuosly evolving datasets. Dimensionality reduction techniques, such as t-SNE and UMAP, have been widely used to transform high-dimensional features into more discriminative, low-dimensional representations. These transformations improve the effectiveness of retrieval systems by not only preserving but also enhancing the underlying structure of the data. However, their transductive nature requires access to the entire dataset during the reduction process, limiting their use in dynamic environments where data is constantly added. In this paper, we propose ISSDiR, a self-supervised, inductive dimensionality reduction method that generalizes to unseen data, offering a practical solution for continuously expanding datasets. Our approach integrates neural networks-based feature extraction with clustering-based pseudo-labels and introduces a hybrid loss function that combines cross-entropy and constrastive loss, weighted by cluster distances. Extensive experiments demonstrate the competitive performance of the proposed method in multiple datasets. This indicates its potential to contribute to the field of image retrieval by introducing a novel inductive approach specifically designed for dimensionality reduction in retrieval tasks.-
Descrição: dc.descriptionDepartment of Statistics Applied Mathematics and Computing (DEMAC) State University of São Paulo (UNESP)-
Descrição: dc.descriptionDepartment of Statistics Applied Mathematics and Computing (DEMAC) State University of São Paulo (UNESP)-
Formato: dc.format383-391-
Idioma: dc.languageen-
Relação: dc.relationProceedings of the International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications-
???dc.source???: dc.sourceScopus-
Palavras-chave: dc.subjectContent-Based Image Retrieval-
Palavras-chave: dc.subjectDimensionality Reduction-
Palavras-chave: dc.subjectNeural Networks-
Palavras-chave: dc.subjectSelf-Supervised Learning-
Título: dc.titleInductive Self-Supervised Dimensionality Reduction for Image Retrieval-
Tipo de arquivo: dc.typeaula digital-
Aparece nas coleções:Repositório Institucional - Unesp

Não existem arquivos associados a este item.