Research map / Image generation / Fast diffusion sampling
DMGD: Train-Free Dataset Distillation with Semantic-Distribution Matching in Diffusion Models
Qichao Wang, Yunhong Lu, Hengyuan Cao, Junyi Zhang, Min Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add semantic matching (dynamic soft-label guidance) to DiT (ablation) | DMGD · ImageNet-Woof IPC-50 | top-1 accuracy: 49.3 -> 59.3 | Improved |
| add semantic matching to DiT (ablation) | DMGD · ImageNet-Woof IPC-10 | top-1 accuracy: 34.7 -> 38.9 | Improved |
| add distribution matching (OT guidance) to DiT (ablation) | DMGD · ImageNet-Woof IPC-10 | top-1 accuracy: 34.7 -> 41.6 | Improved |
| add distribution matching to DiT (ablation) | DMGD · ImageNet-Woof IPC-50 | top-1 accuracy: 49.3 -> 56.8 | Improved |
| add distribution matching to semantic matching (ablation) | DMGD · ImageNet-Woof IPC-50 | top-1 accuracy: 59.3 -> 60.1 | Improved |
| add semantic matching to distribution matching (ablation) | DMGD · ImageNet-Woof IPC-50 | top-1 accuracy: 56.8 -> 60.1 | Improved |
| add semantic matching to distribution matching (ablation) | DMGD · ImageNet-Woof IPC-10 | top-1 accuracy: 41.6 -> 40.8 | Hurt |
| add distribution matching to semantic matching (ablation) | DMGD · ImageNet-Nette IPC-10 | top-1 accuracy: 67.1 -> 68.4 | Improved |
| add distribution matching to semantic matching (ablation) | DMGD · ImageNet-Nette IPC-50 | top-1 accuracy: 79.7 -> 80.6 | Improved |
| plug DMGD into MiniMax | MiniMax · ImageNet-Nette IPC-10 | top-1 accuracy: 62.0 -> 68.7 | Improved |
| plug DMGD into DiT | DiT · ImageNet-Woof IPC-50 | top-1 accuracy: 49.3 -> 60.1 | Improved |
| use DMGD instead of MGD3 on Woof IPC-50 | DiT + DMGD · ImageNet-Woof IPC-50 | top-1 accuracy: 56.5 -> 60.1 | Improved |
Closest papers by meaning
- D^4: Dataset Distillation via Disentangled Diffusion Model (CVPR 2024 · similarity 0.90)
- $\mathbf{D^{3}S^{2}}$: Diffusion-Guided Dataset Distillation for Semantic Segmentation (NeurIPS 2026 · similarity 0.90)
- MGD$^3$ : Mode-Guided Dataset Distillation using Diffusion Models (ICML 2025 · similarity 0.89)
- DKDM: Data-Free Knowledge Distillation for Diffusion Models with Any Architecture (CVPR 2025 · similarity 0.89)
- Unlocking Dataset Distillation with Diffusion Models (NeurIPS 2025 · similarity 0.89)
- Diffusion Model for Dense Matching (ICLR 2024 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
