Research map / Image generation / Fast diffusion sampling
An Adaptive Sampling Framework for Diffusion-based Dataset Distillation with High Fidelity and Diversity
Sunbeom Jeong, Sehwan Kim, Hyeonggeun Han, Hyungjun Joo, Sangwoo Hong, Jungwoo Lee
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 |
|---|---|---|---|
| use adaptive sampling with repulsion regularization instead of D4M | diffusion DD · ImageNet-1K IPC=50 (ResNet-18) | accuracy (%): 55.2 -> 61.4 | Improved |
| use proposed method instead of D4M | diffusion DD · ImageNet-1K IPC=10 (ResNet-18) | accuracy (%): 27.9 -> 46.6 | Improved |
| use proposed method instead of D4M | diffusion DD · ImageWoof IPC=50 (ResNet-18) | accuracy (%): 71.6 -> 72.6 | Improved |
| use proposed method instead of D4M | diffusion DD · ImageNet-100 IPC=10 (ResNet-18) | accuracy (%): 30.2 -> 39.0 | Improved |
| use proposed method instead of D4M with ResNet-101 | diffusion DD · ImageNet-1K IPC=50 (ResNet-101) | accuracy (%): 63.4 -> 67.8 | Improved |
| use proposed unaligned LDM instead of MGD3 aligned DiT | diffusion DD · ImageNet IPC=50 | accuracy (%): 60.2 -> 61.4 | Improved |
| use proposed method instead of Minimax | diffusion DD · Synthetic image quality | FID: 57.9 -> 48.7 | Improved |
| use proposed method instead of Minimax | diffusion DD · Synthetic image quality | Precision: 0.88 -> 0.84 | Hurt |
| use proposed method instead of D4M | diffusion DD · Synthetic image quality | Recall: 0.14 -> 0.45 | Improved |
| drop latent prototype generation (proposed vs D4M) | diffusion DD · Distillation cost | GPU memory (GB): 6.4 -> 2.6 | Improved |
| training-free proposed method instead of Minimax fine-tuning | diffusion DD · Distillation cost | generation time per image (s): 10.1 -> 2.7 | Improved |
| use proposed method instead of RDED on ImageNette IPC=10 with ResNet-101 | diffusion DD · ImageNette IPC=10 (ResNet-101) | accuracy (%): 54.0 -> 60.4 | Improved |
Closest papers by meaning
- Taming Diffusion for Dataset Distillation with High Representativeness (ICML 2025 · similarity 0.93)
- Unlocking Dataset Distillation with Diffusion Models (NeurIPS 2025 · similarity 0.93)
- MGD$^3$ : Mode-Guided Dataset Distillation using Diffusion Models (ICML 2025 · similarity 0.92)
- D^4: Dataset Distillation via Disentangled Diffusion Model (CVPR 2024 · similarity 0.92)
- Efficient Dataset Distillation via Minimax Diffusion (CVPR 2024 · similarity 0.91)
- Large Scale Dataset Distillation with Domain Shift (ICML 2024 · similarity 0.90)
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.
