Guardado en:
Detalles Bibliográficos
Autores principales: Collins, Emma, wong, Myungseo, Yun, Kim, Kingston, Finn, Satou, Hana
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2505.15241
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917123117285376
author Collins, Emma
wong, Myungseo
Yun, Kim
Kingston, Finn
Satou, Hana
author_facet Collins, Emma
wong, Myungseo
Yun, Kim
Kingston, Finn
Satou, Hana
contents Despite progress in geometry-aware domain adaptation, current methods such as GAMA still suffer from two unresolved issues: (1) insufficient disentanglement of task-relevant and task-irrelevant manifold dimensions, and (2) rigid perturbation schemes that ignore per-class alignment asymmetries. To address this, we propose GAMA++, a novel framework that introduces (i) latent space disentanglement to isolate label-consistent manifold directions from nuisance factors, and (ii) an adaptive contrastive perturbation strategy that tailors both on- and off-manifold exploration to class-specific manifold curvature and alignment discrepancy. We further propose a cross-domain contrastive consistency loss that encourages local semantic clusters to align while preserving intra-domain diversity. Our method achieves state-of-the-art results on DomainNet, Office-Home, and VisDA benchmarks under both standard and few-shot settings, with notable improvements in class-level alignment fidelity and boundary robustness. GAMA++ sets a new standard for semantic geometry alignment in transfer learning.
format Preprint
id arxiv_https___arxiv_org_abs_2505_15241
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Disentangled Geometric Alignment with Adaptive Contrastive Perturbation for Reliable Domain Transfer
Collins, Emma
wong, Myungseo
Yun, Kim
Kingston, Finn
Satou, Hana
Computer Vision and Pattern Recognition
Despite progress in geometry-aware domain adaptation, current methods such as GAMA still suffer from two unresolved issues: (1) insufficient disentanglement of task-relevant and task-irrelevant manifold dimensions, and (2) rigid perturbation schemes that ignore per-class alignment asymmetries. To address this, we propose GAMA++, a novel framework that introduces (i) latent space disentanglement to isolate label-consistent manifold directions from nuisance factors, and (ii) an adaptive contrastive perturbation strategy that tailors both on- and off-manifold exploration to class-specific manifold curvature and alignment discrepancy. We further propose a cross-domain contrastive consistency loss that encourages local semantic clusters to align while preserving intra-domain diversity. Our method achieves state-of-the-art results on DomainNet, Office-Home, and VisDA benchmarks under both standard and few-shot settings, with notable improvements in class-level alignment fidelity and boundary robustness. GAMA++ sets a new standard for semantic geometry alignment in transfer learning.
title Disentangled Geometric Alignment with Adaptive Contrastive Perturbation for Reliable Domain Transfer
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.15241