Research map / Label-efficient and robust learning / Recognition and long-tail
Balanced Dataset Distillation via Modeling Multiple Visual Pattern Distribution
Guanghui Shi, Xuefeng Liang, Qixiang Wen
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 |
|---|---|---|---|
| select samples from both pattern centers and margins in balance (BPS) instead of unbalanced selection (ablation) | BPS coreset · ImageNet-1K IPC=50 | accuracy and MMD | Improved |
| model multiple visual patterns per class and balance selection (BPS) vs CCFS | BPS · CIFAR-10 IPC=50 | accuracy (%): 67.2 -> 77.1 | Improved |
| model-agnostic coreset selection for cross-architecture evaluation | BPS · ImageNet-1K IPC=10, ViT-S | accuracy (%): 24.0 -> 30.2 | Improved |
| raise candidate ratio K from 10% to 20% (ablation) | BPS · ImageNet-1K IPC=10 | accuracy (%): 51.9 -> 52.4 | Improved |
| raise candidate ratio K from 10% to 20% (ablation) | BPS · CIFAR-100 IPC=10 | accuracy (%): 57.7 -> 57.1 | Hurt |
| raise candidate ratio K from 20% to 30% (ablation) | BPS · ImageNet-1K IPC=10 | accuracy (%): 52.4 -> 51.6 | Hurt |
| use intermediate contrastive temperature tau1=0.5 instead of small or large values (ablation) | BPS Stage 1 · CIFAR-10/CIFAR-100 IPC=10 | accuracy | Improved |
| use intermediate distillation temperature tau2=20 instead of small or large values (ablation) | BPS Stage 3 · CIFAR-10/CIFAR-100 IPC=10 | accuracy | Improved |
| reuse one Stage 1 optimization across IPC budgets (Optimize-Once-for-All-IPCs) | BPS · CIFAR-100 distillation time | time cost | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DELT — DELT: A Simple Diversity-driven EarlyLate Training for Dataset Distillation (CVPR 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Rethinking Dataset Distillation for Classification: Do Distilled Sets Outperform Coresets? (NeurIPS 2026 · similarity 0.90)
- Curriculum Coarse-to-Fine Selection for High-IPC Dataset Distillation (CVPR 2025 · similarity 0.88)
- D^4: Dataset Distillation via Disentangled Diffusion Model (CVPR 2024 · similarity 0.88)
- Enhancing Dataset Distillation via Non-Critical Region Refinement (CVPR 2025 · similarity 0.87)
- Large Scale Dataset Distillation with Domain Shift (ICML 2024 · similarity 0.87)
- Emphasizing Discriminative Features for Dataset Distillation in Complex Scenarios (CVPR 2025 · similarity 0.87)
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.
