D$^4$M: Dataset Distillation via Disentangled Diffusion Model
This work addresses the computational cost and cross-architecture performance decline in dataset distillation, benefiting researchers and practitioners in machine learning by enabling faster and more versatile training.
The paper tackles the problem of dataset distillation by proposing D$^4$M, an efficient framework that uses a disentangled diffusion model to create synthetic datasets independent of matching architectures, achieving superior performance and robust generalization compared to state-of-the-art methods.
Dataset distillation offers a lightweight synthetic dataset for fast network training with promising test accuracy. To imitate the performance of the original dataset, most approaches employ bi-level optimization and the distillation space relies on the matching architecture. Nevertheless, these approaches either suffer significant computational costs on large-scale datasets or experience performance decline on cross-architectures. We advocate for designing an economical dataset distillation framework that is independent of the matching architectures. With empirical observations, we argue that constraining the consistency of the real and synthetic image spaces will enhance the cross-architecture generalization. Motivated by this, we introduce Dataset Distillation via Disentangled Diffusion Model (D$^4$M), an efficient framework for dataset distillation. Compared to architecture-dependent methods, D$^4$M employs latent diffusion model to guarantee consistency and incorporates label information into category prototypes. The distilled datasets are versatile, eliminating the need for repeated generation of distinct datasets for various architectures. Through comprehensive experiments, D$^4$M demonstrates superior performance and robust generalization, surpassing the SOTA methods across most aspects.