Diversity-Driven Generative Dataset Distillation Based on Diffusion Model with Self-Adaptive Memory

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Li, Mingzhuo, Li, Guang, Mao, Jiafeng, Ogawa, Takahiro, Haseyama, Miki
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916758965714944
author Li, Mingzhuo
Li, Guang
Mao, Jiafeng
Ogawa, Takahiro
Haseyama, Miki
author_facet Li, Mingzhuo
Li, Guang
Mao, Jiafeng
Ogawa, Takahiro
Haseyama, Miki
contents Dataset distillation enables the training of deep neural networks with comparable performance in significantly reduced time by compressing large datasets into small and representative ones. Although the introduction of generative models has made great achievements in this field, the distributions of their distilled datasets are not diverse enough to represent the original ones, leading to a decrease in downstream validation accuracy. In this paper, we present a diversity-driven generative dataset distillation method based on a diffusion model to solve this problem. We introduce self-adaptive memory to align the distribution between distilled and real datasets, assessing the representativeness. The degree of alignment leads the diffusion model to generate more diverse datasets during the distillation process. Extensive experiments show that our method outperforms existing state-of-the-art methods in most situations, proving its ability to tackle dataset distillation tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2505_19469
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Diversity-Driven Generative Dataset Distillation Based on Diffusion Model with Self-Adaptive Memory
Li, Mingzhuo
Li, Guang
Mao, Jiafeng
Ogawa, Takahiro
Haseyama, Miki
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Dataset distillation enables the training of deep neural networks with comparable performance in significantly reduced time by compressing large datasets into small and representative ones. Although the introduction of generative models has made great achievements in this field, the distributions of their distilled datasets are not diverse enough to represent the original ones, leading to a decrease in downstream validation accuracy. In this paper, we present a diversity-driven generative dataset distillation method based on a diffusion model to solve this problem. We introduce self-adaptive memory to align the distribution between distilled and real datasets, assessing the representativeness. The degree of alignment leads the diffusion model to generate more diverse datasets during the distillation process. Extensive experiments show that our method outperforms existing state-of-the-art methods in most situations, proving its ability to tackle dataset distillation tasks.
title Diversity-Driven Generative Dataset Distillation Based on Diffusion Model with Self-Adaptive Memory
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2505.19469