DMGD: Train-Free Dataset Distillation with Semantic-Distribution Matching in Diffusion Models

TL;DR

The paper introduces DMGD, a training-free dataset distillation method using diffusion models with semantic and distribution matching, achieving state-of-the-art results efficiently.

Contribution

Proposes a novel training-free diffusion-based dataset distillation framework with semantic and distribution matching, eliminating the need for fine-tuning.

Findings

Outperforms SOTA methods on ImageNet variants with accuracy gains of 2.1%, 5.4%, and 2.4%.

Introduces semantic matching via likelihood optimization without auxiliary classifiers.

Develops efficient strategies for distribution matching with minimal computational overhead.

Abstract

Dataset distillation enables efficient training by distilling the information of large-scale datasets into significantly smaller synthetic datasets. Diffusion based paradigms have emerged in recent years, offering novel perspectives for dataset distillation. However, they typically necessitate additional fine-tuning stages, and effective guidance mechanisms remain underexplored. To address these limitations, we rethink diffusion based dataset distillation and propose a Dual Matching Guided Diffusion (DMGD) framework, centered on efficient training-free guidance. We first establish Semantic Matching via conditional likelihood optimization, eliminating the need for auxiliary classifiers. Furthermore, we propose a dynamic guidance mechanism that enhances the diversity of synthetic data while maintaining semantic alignment. Simultaneously, we introduce an optimal transport (OT) based Distribution Matching approach to further align with the target distribution structure. To ensure efficiency, we develop two enhanced strategies for diffusion based framework: Distribution Approximate Matching and Greedy Progressive Matching. These strategies enable effective distribution matching guidance with minimal computational overhead. Experimental results on ImageNet-Woof, ImageNet-Nette, and ImageNet-1K demonstrate that our training-free approach achieves significant improvements, outperforming state-of-the-art (SOTA) methods requiring additional fine-tuning by average accuracy gains of 2.1%, 5.4%, and 2.4%, respectively.