Research map / Architectures and efficiency / Pruning and efficient tuning
Explore and Establish Synergistic Effects Between Weight Pruning and Coreset Selection in Neural Network Training
Weilin Wan, Fan Yi, Weizhong Zhang, Quan Zhou, Cheng Jin
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 30% weight pruning to coreset selection (SWaST-cut) | coreset-only training · CIFAR-100 1% coreset, ResNet-18 | accuracy (%): 36.42 -> 39.55 | Improved |
| add pruning to coreset selection (SWaST-cut, 30% prune) | coreset-only · CIFAR-10 1% coreset, ResNet-101 | accuracy (%): 49.72 -> 67.55 (+17.83) | Improved |
| add pruning to coreset selection (SWaST-trim 90%) | coreset-only · CIFAR-10 10% coreset, ResNet-18 | accuracy (%): 92.20 -> 92.58 | No clear effect |
| shrink coreset size (10% -> 1%) (ablation) | SWaST · CIFAR-10, ResNet-101 | accuracy gain over coreset-only: +1.47 -> +16.51 | Improved |
| aggressive 90% backbone pruning with SWaST-cut (ablation) | SWaST-cut · CIFAR-10 10% coreset, ResNet-18 | accuracy (%): 92.20 -> 91.53 (-0.67) | Hurt |
| aggressive 90% backbone pruning with SWaST-cut (ablation) | SWaST-cut · CIFAR-100 10% coreset, ResNet-18 | accuracy (%): 70.97 -> 67.98 (-2.99) | Hurt |
| add coreset selection to pruning-only training | SWaST-cut · CIFAR-10 at 90% prune rate | accuracy (%): +3.33 to 93.15 | Improved |
| add pruning to coreset selection with label noise | SWaST-cut · noisy CIFAR, EL2N coreset | noise ratio in coreset: 10.62% reduction | Improved |
| add pruning | SWaST · coreset training overfitting | test minus val loss | Improved |
| add state preservation (KL consistency, lambda 0.1) (ablation) | SWaST-cut · aggressive pruning | training stability | Improved |
Closest papers by meaning
- Pruning Neural Networks via Coresets and Convex Geometry: Towards No Assumptions (NeurIPS 2022 · similarity 0.89)
- $\mathbb{D}^2$ Pruning: Message Passing for Balancing Diversity & Difficulty in Data Pruning (ICLR 2024 · similarity 0.89)
- Stable Coresets via Posterior Sampling: Aligning Induced and Full Loss Landscapes (NeurIPS 2025 · similarity 0.88)
- Refined Coreset Selection: Towards Minimal Coreset Size under Model Performance Constraints (ICML 2024 · similarity 0.87)
- The Impact of Coreset Selection on Spurious Correlations and Group Robustness (NeurIPS 2025 · similarity 0.87)
- Data Pruning by Information Maximization (ICLR 2025 · similarity 0.86)
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.
