Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Quintana, Gonzalo Iñaki, Vancamberg, Laurence, Jugnon, Vincent, Desolneux, Agnès, Mougeot, Mathilde
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2502.00052
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866929693464199168
author Quintana, Gonzalo Iñaki
Vancamberg, Laurence
Jugnon, Vincent
Desolneux, Agnès
Mougeot, Mathilde
author_facet Quintana, Gonzalo Iñaki
Vancamberg, Laurence
Jugnon, Vincent
Desolneux, Agnès
Mougeot, Mathilde
contents This work studies the relationship between Contrastive Learning and Domain Adaptation from a theoretical perspective. The two standard contrastive losses, NT-Xent loss (Self-supervised) and Supervised Contrastive loss, are related to the Class-wise Mean Maximum Discrepancy (CMMD), a dissimilarity measure widely used for Domain Adaptation. Our work shows that minimizing the contrastive losses decreases the CMMD and simultaneously improves class-separability, laying the theoretical groundwork for the use of Contrastive Learning in the context of Domain Adaptation. Due to the relevance of Domain Adaptation in medical imaging, we focused the experiments on mammography images. Extensive experiments on three mammography datasets - synthetic patches, clinical (real) patches, and clinical (real) images - show improved Domain Adaptation, class-separability, and classification performance, when minimizing the Supervised Contrastive loss.
format Preprint
id arxiv_https___arxiv_org_abs_2502_00052
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Bridging Contrastive Learning and Domain Adaptation: Theoretical Perspective and Practical Application
Quintana, Gonzalo Iñaki
Vancamberg, Laurence
Jugnon, Vincent
Desolneux, Agnès
Mougeot, Mathilde
Machine Learning
Artificial Intelligence
This work studies the relationship between Contrastive Learning and Domain Adaptation from a theoretical perspective. The two standard contrastive losses, NT-Xent loss (Self-supervised) and Supervised Contrastive loss, are related to the Class-wise Mean Maximum Discrepancy (CMMD), a dissimilarity measure widely used for Domain Adaptation. Our work shows that minimizing the contrastive losses decreases the CMMD and simultaneously improves class-separability, laying the theoretical groundwork for the use of Contrastive Learning in the context of Domain Adaptation. Due to the relevance of Domain Adaptation in medical imaging, we focused the experiments on mammography images. Extensive experiments on three mammography datasets - synthetic patches, clinical (real) patches, and clinical (real) images - show improved Domain Adaptation, class-separability, and classification performance, when minimizing the Supervised Contrastive loss.
title Bridging Contrastive Learning and Domain Adaptation: Theoretical Perspective and Practical Application
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2502.00052