Contrastive Loss Based on Contextual Similarity for Image Classification

Registro completo de metadados
MetadadosDescriçãoIdioma
Autor(es): dc.contributorUniversidade Estadual Paulista (UNESP)-
Autor(es): dc.contributorUniversité du Quebec en Outaouais-
Autor(es): dc.creatorValem, Lucas Pascotti-
Autor(es): dc.creatorPedronette, Daniel Carlos Guimarães-
Autor(es): dc.creatorAllili, Mohand Said-
Data de aceite: dc.date.accessioned2025-08-21T20:37:09Z-
Data de disponibilização: dc.date.available2025-08-21T20:37:09Z-
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/10.1007/978-3-031-77392-1_5-
Fonte completa do material: dc.identifierhttps://hdl.handle.net/11449/308911-
Fonte: dc.identifier.urihttp://educapes.capes.gov.br/handle/11449/308911-
Descrição: dc.descriptionContrastive learning has been extensively exploited in self-supervised and supervised learning due to its effectiveness in learning representations that distinguish between similar and dissimilar images. It offers a robust alternative to cross-entropy by yielding more semantically meaningful image embeddings. However, most contrastive losses rely on pairwise measures to assess the similarity between elements, ignoring more general neighborhood information that can be leveraged to enhance model robustness and generalization. In this paper, we propose the Contextual Contrastive Loss (CCL) to replace pairwise image comparison by introducing a new contextual similarity measure using neighboring elements. The CCL yields a more semantically meaningful image embedding ensuring better separability of classes in the latent space. Experimental evaluation on three datasets (Food101, MiniImageNet, and CIFAR-100) has shown that CCL yields superior results by achieving up to 10.76% relative gains in classification accuracy, particularly for fewer training epochs and limited training data. This demonstrates the potential of our approach, especially in resource-constrained scenarios.-
Descrição: dc.descriptionPetrobras-
Descrição: dc.descriptionSão Paulo State University (UNESP), SP-
Descrição: dc.descriptionUniversité du Quebec en Outaouais-
Descrição: dc.descriptionSão Paulo State University (UNESP), SP-
Formato: dc.format58-69-
Idioma: dc.languageen-
Relação: dc.relationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)-
???dc.source???: dc.sourceScopus-
Palavras-chave: dc.subjectContrastive Learning-
Palavras-chave: dc.subjectImage Classification-
Título: dc.titleContrastive Loss Based on Contextual Similarity for Image Classification-
Tipo de arquivo: dc.typeaula digital-
Aparece nas coleções:Repositório Institucional - Unesp

Não existem arquivos associados a este item.