cs.CVNov 12, 2026

Disentangled Embedding Optimization for Diffusion-Based Dataset Distillation

Authors: Hongcheng Li, Yucan Zhou, Yibin Wang, Xiaoyan Gu, Bo Li, Weiping Wang

Organizations: Institute of Information Engineering, Chinese Academy of Sciences & School of Cyber Security, University of Chinese Academy of Sciences & State Key Laboratory of Cyberspace Security Defense, Beijing, China · College of Intelligence and Computing, Tianjin University, Tianjin, China · Institute of Information Engineering, Chinese Academy of Sciences State Key Laboratory of Cyberspace Security Defense, Beijing, China · Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China

Abstract

Dataset distillation aims to compress large-scale datasets into compact synthetic subsets that enable efficient model training with comparable performance to the original data. Recent advances show that diffusion models are highly effective for this task due to their strong generative capability. Existing diffusion-based approaches typically perform class-wise clustering and use the obtained cluster prototypes as diffusion inputs to generate distilled samples. However, the uniformly distributed original samples in a category inevitably lead to entanglement near cluster boundaries, pulling the cluster prototypes closer and reducing the inter-cluster separation. Consequently, the diffusion model receives overly similar latent inputs, resulting in synthetic data with limited diversity. To address this issue, we propose Disentangled Embedding Optimization (DEO), a simple yet effective framework that replaces static clustering with trainable embeddings. These embeddings are optimized to match the Gaussian statistics (i.e., mean and covariance) of the original data distribution, while a diversity regularizer is designed to encourage separation among these embeddings to improve the intra-class variability. Extensive experiments demonstrate that our proposed DEO can generate more diversified and representative distilled data than existing cluster-based diffusion methods, achieving state-of-the-art performance and superior generalization across multiple datasets.

Figures & tables

Explore similar work

CardsList
  1. IMS3: Breaking Distributional Aggregation in Diffusion-Based Dataset Distillation

    Mar 14, 2026Chenru Wang, Yunyi Chen, Zijun Yang +2Diffusion-Based Dataset DistillationDataset Distillation

  2. Towards Consistent and Efficient Dataset Distillation via Diffusion-Driven Selection

    Dec 13, 2024Xinhao Zhong, Shuoyang Sun, Zhaoyang Xu +4Diffusion-Based Dataset DistillationDataset Distillation

  3. Dataset Distillation Based on Saliency-Driven Prototype Alignment

    Jul 28, 2026Yawen Zou, Wenqi Cai, Guang Li +3Diffusion-Based Dataset DistillationPrototype Learning