Research map / Architectures and efficiency / Pruning and efficient tuning
Revisiting Random Channel Pruning for Neural Network Compression
Check 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 |
|---|---|---|---|
| Randomly search channel configurations instead of relying on a fixed pruning configuration | ResNet50 · ImageNet | Top-1 error (%): 25.39 -> 24.87 | Helped |
| Randomly search channel configurations instead of relying on a fixed pruning configuration | MobileNetV2 · ImageNet2012 | Top-1 error (%): 28.55 -> 29.10 | Hurt |
| Randomly search channel configurations instead of using Hinge pruning | ResNet56 · CIFAR10 | Top-1 error (%): 6.31 -> 6.52 | Hurt |
| Use advanced channel importance criteria rather than simple L1/L2 norms | Random channel pruning across ImageNet and CIFAR benchmarks · ImageNet and CIFAR10/CIFAR100 | Top-1 error (%) | No clear effect |
| Increase the random-sample population from 20 to 1000 (ablation) | Random Pruning · ImageNet, ResNet18 | Top-1 error | Helped |
| Increase fine-tuning from 25 to 100 epochs (ablation) | Random Pruning · ImageNet, ResNet50 | Top-1 error (%): 0.75% | Helped |
| Increase fine-tuning from 25 to 100 epochs (ablation) | Random Pruning · ImageNet, ResNet18 | Top-1 error rate: 0.97% | Helped |
Closest papers by meaning
- Network Pruning via Performance Maximization (CVPR 2021 · similarity 0.91)
- CHIP: CHannel Independence-based Pruning for Compact Neural Networks (NeurIPS 2021 · similarity 0.91)
- Discrimination-aware Channel Pruning for Deep Neural Networks (NeurIPS 2018 · similarity 0.91)
- Rethinking the Pruning Criteria for Convolutional Neural Network (NeurIPS 2021 · similarity 0.90)
- Pruning from Scratch (AAAI 2020 · similarity 0.90)
- EPSD: Early Pruning with Self-Distillation for Efficient Model Compression (AAAI 2024 · similarity 0.88)
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.