Research map / Architectures and efficiency / Neural architecture search
MetaPruning: Meta Learning for Automatic Neural Network Channel Pruning
Zechun Liu, Haoyuan Mu, Xiangyu Zhang, Zichao Guo, Xin Yang, Kwang-Ting Cheng, Jian Sun
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 weight prediction (two FC layers) to the PruningNet (ablation) | MetaPruning MobileNetV1 · ImageNet, under 45M FLOPs | top-1 accuracy (%): 55.3 -> 1.9% higher | Improved |
| add weight prediction to the PruningNet (ablation) | PruningNet · ImageNet uniformly pruned MobileNetV1 | PruningNet accuracy: 10% lower accuracy without | Improved |
| use MetaPruning instead of uniform pruning | MobileNetV1 0.25x · ImageNet, FLOPs constraint | top-1 accuracy (%): 50.6 -> 57.2 | Improved |
| use MetaPruning instead of uniform pruning | MobileNetV2 at 43M FLOPs · ImageNet, FLOPs constraint | top-1 accuracy (%): 54.6 -> 58.3 | Improved |
| use MetaPruning instead of uniform pruning | ResNet-50 at about 1G FLOPs · ImageNet, FLOPs constraint | top-1 accuracy (%): 72.0 -> 73.4 | Improved |
| use MetaPruning instead of uniform pruning under latency | MobileNetV1 0.25x · ImageNet, latency constraint | top-1 accuracy (%): 50.6 -> 59.6 | Improved |
| use MetaPruning instead of uniform pruning under latency | MobileNetV2 0.35x · ImageNet, latency constraint | top-1 accuracy (%): 54.6 -> 64.5 | Improved |
| use MetaPruning instead of AMC | MobileNetV1 ~285M FLOPs · ImageNet | top-1 accuracy (%): 70.5 -> 70.6 | Improved |
| use MetaPruning instead of AMC | MobileNetV2 220M FLOPs · ImageNet | top-1 accuracy (%): 70.8 -> 71.2 | Improved |
| use MetaPruning instead of ThiNet | ResNet-50 ~2G FLOPs · ImageNet | top-1 accuracy (%): 74.7 -> 75.4 | Improved |
| prune shortcut channels jointly with middle channels | MobileNetV2 · ImageNet, FLOPs constraint | top-1 accuracy (%): 54.6 -> 58.3 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- NetAdapt 284M — NetAdapt: Platform-Aware Neural Network Adaptation for Mobile Applications (ECCV 2018) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Pruning from Scratch (AAAI 2020 · similarity 0.91)
- Adaptive Search-and-Training for Robust and Efficient Network Pruning (TPAMI 2023 · similarity 0.90)
- Network Pruning via Transformable Architecture Search (NeurIPS 2019 · similarity 0.90)
- Network Pruning via Performance Maximization (CVPR 2021 · similarity 0.90)
- Revisiting Random Channel Pruning for Neural Network Compression (CVPR 2022 · similarity 0.89)
- BilevelPruning: Unified Dynamic and Static Channel Pruning for Convolutional Neural Networks (CVPR 2024 · similarity 0.89)
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.
