Research map / Architectures and efficiency / CNN design
SCOP: Scientific Control for Reliable Neural Network Pruning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 knockoff features as control group (ablation) | SCOP · CIFAR-10 ResNet-56 | error: 6.83 -> 6.36 | Helped |
| add knockoff features as control group (ablation) | SCOP · CIFAR-10 ResNet-32 | error: 8.23 -> 7.78 | Helped |
| use knockoff data instead of Gaussian noise as auxiliary data (ablation) | SCOP · CIFAR-10 ResNet-56 | error: 6.81 -> 6.36 | Helped |
| use knockoff data instead of random samples as auxiliary data (ablation) | SCOP · CIFAR-10 ResNet-56 | error: 6.77 -> 6.36 | Helped |
| add second-order knockoff bias terms (ablation) | SCOP · CIFAR-10 ResNet-56/32 | error: 6.47 -> 6.36 | No clear effect |
| SCOP vs prior filter pruning criteria | ResNet-56 · CIFAR-10 | error: 6.83 -> 6.36 (vs HRank) | Helped |
| SCOP vs prior pruning | ResNet-50 · ImageNet | top-1 error: 25.50 -> 24.74 (vs Taylor) | Helped |
| SCOP vs SFP at similar FLOPs | ResNet-20 · CIFAR-10 | error: 9.17 vs 9.25 | Hurt |
| increase pruning rate (ablation) | SCOP ResNet-56 · CIFAR-10 | accuracy | No clear effect |
Closest papers by meaning
- Pruning from Scratch (AAAI 2020 · similarity 0.89)
- Importance Estimation for Neural Network Pruning (CVPR 2019 · similarity 0.88)
- Provable Filter Pruning for Efficient Neural Networks (ICLR 2020 · similarity 0.87)
- Pruning Neural Networks at Initialization: Why Are We Missing the Mark? (ICLR 2021 · similarity 0.87)
- HRank: Filter Pruning Using High-Rank Feature Map (CVPR 2020 · similarity 0.87)
- Learning to Prune Deep Neural Networks via Layer-wise Optimal Brain Surgeon (NeurIPS 2017 · 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.