Optimizing Distributional Geometry Alignment with Optimal Transport for Generative Dataset Distillation

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Cui, Xiao, Qin, Yulei, Zhou, Wengang, Li, Hongsheng, Li, Houqiang
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915644742565888
author Cui, Xiao
Qin, Yulei
Zhou, Wengang
Li, Hongsheng
Li, Houqiang
author_facet Cui, Xiao
Qin, Yulei
Zhou, Wengang
Li, Hongsheng
Li, Houqiang
contents Dataset distillation seeks to synthesize a compact distilled dataset, enabling models trained on it to achieve performance comparable to models trained on the full dataset. Recent methods for large-scale datasets focus on matching global distributional statistics (e.g., mean and variance), but overlook critical instance-level characteristics and intraclass variations, leading to suboptimal generalization. We address this limitation by reformulating dataset distillation as an Optimal Transport (OT) distance minimization problem, enabling fine-grained alignment at both global and instance levels throughout the pipeline. OT offers a geometrically faithful framework for distribution matching. It effectively preserves local modes, intra-class patterns, and fine-grained variations that characterize the geometry of complex, high-dimensional distributions. Our method comprises three components tailored for preserving distributional geometry: (1) OT-guided diffusion sampling, which aligns latent distributions of real and distilled images; (2) label-image-aligned soft relabeling, which adapts label distributions based on the complexity of distilled image distributions; and (3) OT-based logit matching, which aligns the output of student models with soft-label distributions. Extensive experiments across diverse architectures and large-scale datasets demonstrate that our method consistently outperforms state-of-the-art approaches in an efficient manner, achieving at least 4% accuracy improvement under IPC=10 settings for each architecture on ImageNet-1K.
format Preprint
id arxiv_https___arxiv_org_abs_2512_00308
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Optimizing Distributional Geometry Alignment with Optimal Transport for Generative Dataset Distillation
Cui, Xiao
Qin, Yulei
Zhou, Wengang
Li, Hongsheng
Li, Houqiang
Computer Vision and Pattern Recognition
Dataset distillation seeks to synthesize a compact distilled dataset, enabling models trained on it to achieve performance comparable to models trained on the full dataset. Recent methods for large-scale datasets focus on matching global distributional statistics (e.g., mean and variance), but overlook critical instance-level characteristics and intraclass variations, leading to suboptimal generalization. We address this limitation by reformulating dataset distillation as an Optimal Transport (OT) distance minimization problem, enabling fine-grained alignment at both global and instance levels throughout the pipeline. OT offers a geometrically faithful framework for distribution matching. It effectively preserves local modes, intra-class patterns, and fine-grained variations that characterize the geometry of complex, high-dimensional distributions. Our method comprises three components tailored for preserving distributional geometry: (1) OT-guided diffusion sampling, which aligns latent distributions of real and distilled images; (2) label-image-aligned soft relabeling, which adapts label distributions based on the complexity of distilled image distributions; and (3) OT-based logit matching, which aligns the output of student models with soft-label distributions. Extensive experiments across diverse architectures and large-scale datasets demonstrate that our method consistently outperforms state-of-the-art approaches in an efficient manner, achieving at least 4% accuracy improvement under IPC=10 settings for each architecture on ImageNet-1K.
title Optimizing Distributional Geometry Alignment with Optimal Transport for Generative Dataset Distillation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.00308