Research map / Learning theory and algorithms / Kernels and regression theory
Utility Boundary of Dataset Distillation: Scaling and Coverage Laws
Zhengquan Luo, zhiqiang xu
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 |
|---|---|---|---|
| increase distilled set size k (images per class) | DC, DSA, DM, MTT · MNIST/CIFAR-10/100 | generalization gap: decays linearly in 1/sqrt(k), then saturates at a floor | Improved |
| use MTT (trajectory matching) instead of gradient matching | MTT · CIFAR-100 | linearity of scaling fit: R2 0.73 | Hurt |
| increase configuration diversity (covering complexity) at fixed k | DD methods · MNIST/CIFAR-10 | generalization gap: linear in sqrt(Hcov/k) | Hurt |
| test coverage law on harder dataset | coverage law fit · CIFAR-100 | R2 of linear fit: R2 down to about 0.3 for some methods | Mixed |
| scaling law prediction on held-out IPC | scaling law · MNIST/CIFAR-10/100 | prediction MAE: MAE = 0.0214 | Improved |
Closest papers by meaning
- Embarrassingly Simple Dataset Distillation (ICLR 2024 · similarity 0.90)
- Towards Lossless Dataset Distillation via Difficulty-Aligned Trajectory Matching (ICLR 2024 · similarity 0.90)
- Rethinking Dataset Distillation for Classification: Do Distilled Sets Outperform Coresets? (NeurIPS 2026 · similarity 0.89)
- D^4: Dataset Distillation via Disentangled Diffusion Model (CVPR 2024 · similarity 0.88)
- SelMatch: Effectively Scaling Up Dataset Distillation via Selection-Based Initialization and Partial Updates by Trajectory Matching (ICML 2024 · similarity 0.88)
- Group Distributionally Robust Dataset Distillation with Risk Minimization (ICLR 2025 · similarity 0.88)
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.
