Research map / Architectures and efficiency / Pruning and efficient tuning
Collaborative Channel Pruning for Deep Networks
Hanyu Peng, Jiaxiang Wu, Shifeng Chen, Junzhou Huang
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 |
|---|---|---|---|
| CCP versus DCP at similar FLOPs reduction, top-5 drop | CCP · ILSVRC-12 ResNet-50 | top-5 accuracy drop: 0.61% -> 0.45% | Improved |
| Add auxiliary classifier (CCP-AC) on ILSVRC-12, top-5 (ablation) | CCP · ILSVRC-12 ResNet-50 | top-5 accuracy drop: 0.45% -> 0.33% | Improved |
| Add auxiliary classifier (CCP-AC) on ILSVRC-12, top-1 (ablation) | CCP · ILSVRC-12 ResNet-50 | top-1 accuracy drop: 0.94% -> 0.83% | Improved |
| Add auxiliary classifier (CCP-AC) on CIFAR-10 (ablation) | CCP · CIFAR-10 ResNet-56 | accuracy drop: 0.04% -> -0.19% | Improved |
| CCP versus DCP-Adapt at 47.0% FLOPs reduction | CCP · CIFAR-10 ResNet-56 | accuracy drop: -0.01% -> 0.04% | Hurt |
| CCP versus DCP at ~50% FLOPs on CIFAR-10 | CCP · CIFAR-10 ResNet-56 | accuracy drop: 0.31% -> 0.08% | Improved |
| CCP versus Channel Pruning (He et al.) on CIFAR-10 | CCP · CIFAR-10 ResNet-56 | accuracy drop: 1.00% -> 0.04% | Improved |
| CCP versus Soft Pruning on ILSVRC-12 top-1 | CCP · ILSVRC-12 ResNet-50 | top-1 accuracy drop: 1.54% -> 0.65% | Improved |
| Raise pruning ratio r from 0.35 to 0.40 on ILSVRC-12 | CCP · ILSVRC-12 ResNet-50 | top-1 accuracy drop: 0.65% -> 0.94% | Hurt |
| Raise pruning ratio r from 0.35 to 0.40 on CIFAR-10 | CCP · CIFAR-10 ResNet-56 | accuracy drop: 0.04% -> 0.08% | Hurt |
| Vary auxiliary classifier coefficient lambda from 0 to 1e-1 (r=0.35) (ablation) | CCP-AC · CIFAR-10 ResNet-56 | accuracy drop | Improved |
| Increase pruning ratio from 0.1 to 0.6 (ablation) | CCP · CIFAR-10 ResNet-56 | accuracy drop | Mixed |
Closest papers by meaning
- Discrimination-aware Channel Pruning for Deep Neural Networks (NeurIPS 2018 · similarity 0.90)
- Discrimination-aware Network Pruning for Deep Model Compression (TPAMI 2021 · similarity 0.89)
- REPrune: Channel Pruning via Kernel Representative Selection (AAAI 2024 · similarity 0.87)
- How Sparse Can We Prune A Deep Network: A Fundamental Limit Perspective (NeurIPS 2024 · similarity 0.86)
- Device-Wise Federated Network Pruning (CVPR 2024 · similarity 0.86)
- BilevelPruning: Unified Dynamic and Static Channel Pruning for Convolutional Neural Networks (CVPR 2024 · 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.
