Research map / Architectures and efficiency / CNN design
Rethinking the Smaller-Norm-Less-Informative Assumption in Channel Pruning of Convolution Layers
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 |
|---|---|---|---|
| Apply ISTA channel pruning to an over-parameterized ConvNet | ConvNet · CIFAR-10 | test accuracy (%): 89.0 -> 89.5 | Helped |
| Prune further from ConvNet model A with ρ=0.002 | ConvNet · CIFAR-10 | test accuracy (%): 89.5 -> 87.6 | Hurt |
| Prune further from ConvNet model B with ρ=0.008 | ConvNet · CIFAR-10 | test accuracy (%): 87.6 -> 86.0 | Hurt |
| Use ISTA channel pruning | ResNet-20 · CIFAR-10 | test accuracy (%): 92.0 -> 90.9 | Hurt |
| Prune further from ResNet-20 model A | ResNet-20 · CIFAR-10 | test accuracy (%): 90.9 -> 88.8 | Hurt |
| Apply ISTA channel pruning and fine-tuning | ResNet-101 · ILSVRC2012 | single-crop Top-1 error (%): 23.6 -> 24.73 | Hurt |
| Apply more aggressive ISTA channel pruning | ResNet-101 · ILSVRC2012 | single-crop Top-1 error (%): 23.6 -> 25.44 | Hurt |
| Apply ISTA channel pruning | segmentation model · MSRA10K | mIOU: 83.4% -> 85.5% | Helped |
| Apply ISTA channel pruning | segmentation model · DUT-Omron | mIOU: 83.2% -> 79.1% | Hurt |
| Apply ISTA channel pruning | segmentation model · Adobe Flickr-portrait | mIOU: 88.6% -> 93.3% | Helped |
| Apply ISTA channel pruning | segmentation model · Adobe Flickr-hp | mIOU: 84.5% -> 89.5% | Helped |
| Apply ISTA channel pruning | segmentation model · COCO-person | mIOU: 84.1% -> 87.5% | Helped |
Closest papers by meaning
- Discrimination-aware Channel Pruning for Deep Neural Networks (NeurIPS 2018 · similarity 0.92)
- Variational Convolutional Neural Network Pruning (CVPR 2019 · similarity 0.92)
- Dynamic Channel Pruning: Feature Boosting and Suppression (ICLR 2019 · similarity 0.91)
- Rethinking the Pruning Criteria for Convolutional Neural Network (NeurIPS 2021 · similarity 0.91)
- BilevelPruning: Unified Dynamic and Static Channel Pruning for Convolutional Neural Networks (CVPR 2024 · similarity 0.91)
- REPrune: Channel Pruning via Kernel Representative Selection (AAAI 2024 · similarity 0.90)
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.