Research map / Architectures and efficiency / Pruning and efficient tuning
You Only Condense Once: Two Rules for Pruning Condensed Datasets
Yang He, Lingao Xiao, Joey Tianyi Zhou
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 |
|---|---|---|---|
| Apply YOCO's low-LBPE ranking and balanced construction when resizing | Condensed CIFAR-10 datasets · CIFAR-10 IPC 10→1,2,5 (Avg.) | accuracy: 41.33 -> 48.30 | Improved |
| Apply YOCO's low-LBPE ranking and balanced construction when resizing | Condensed CIFAR-100 datasets · CIFAR-100 IPC 10→1,2,5 (Avg.) | accuracy: 26.75 -> 30.06 | Improved |
| Apply YOCO's low-LBPE ranking and balanced construction when resizing | Condensed ImageNet-10 datasets · ImageNet-10 IPC 10→1,2,5 (Avg.) | accuracy: 56.66 -> 59.36 | Improved |
| Add LBPE ranking and balanced construction (ablation) | IDC-condensed CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 28.23 -> 42.28 | Improved |
| Add LBPE ranking and balanced construction (ablation) | DREAM-condensed CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 30.87 -> 42.29 | Improved |
| Add LBPE ranking and balanced construction (ablation) | MTT-condensed CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 19.75 -> 22.02 | Improved |
| Add LBPE ranking and balanced construction (ablation) | KIP-condensed CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 14.06 -> 22.24 | Improved |
| Add LBPE ranking to random selection (ablation) | IDC-condensed CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 28.23 -> 39.38 | Improved |
| Add balanced construction to random selection (ablation) | IDC-condensed CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 28.23 -> 30.19 | Improved |
| Add balanced construction to LBPE selection (ablation) | YOCO on CIFAR-10 · CIFAR-10 IPC 10→1,2,5 | accuracy: 39.38 -> 42.28 | Improved |
| Add balanced construction to YOCO selection (ablation) | YOCO on CIFAR-10 · CIFAR-10 IPC 10→1 | accuracy: 37.63 -> 42.28 | Improved |
| Add balanced construction to YOCO selection (ablation) | YOCO on CIFAR-10 · CIFAR-10 IPC 10→2 | accuracy: 42.99 -> 46.67 | Improved |
Closest papers by meaning
- Multisize Dataset Condensation (ICLR 2024 · similarity 0.87)
- Elucidating the Design Space of Dataset Condensation (NeurIPS 2024 · similarity 0.86)
- Lightweight Dataset Pruning without Full Training via Example Difficulty and Prediction Uncertainty (ICML 2025 · similarity 0.85)
- CondTSF: One-line Plugin of Dataset Condensation for Time Series Forecasting (NeurIPS 2024 · similarity 0.84)
- Distilling the Knowledge in Data Pruning (ICML 2025 · similarity 0.84)
- Are Large-scale Soft Labels Necessary for Large-scale Dataset Distillation? (NeurIPS 2024 · similarity 0.84)
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.
