Research map / Architectures and efficiency / CNN design
Filter Pruning via Geometric Median for Deep Convolutional Neural Networks Acceleration
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 |
|---|---|---|---|
| Prune filters nearest to the layer's geometric median rather than selecting by filter norm | ResNet-101 · ILSVRC-2012 | top-1 accuracy drop (%): 2.10 -> 0.05 | Helped |
| Prune 40% using FPGM-only instead of SFP | ResNet-110 · CIFAR-10 | accuracy (%): 93.38 -> 93.73 | Helped |
| Combine norm-based selection with FPGM selection (ablation) | ResNet-110 · CIFAR-10 | accuracy (%): 93.73 -> 93.85 | Helped |
| Combine norm-based selection with FPGM selection (ablation) | ResNet-18 · ILSVRC-2012 | accuracy | No clear effect |
| Use ℓ1 distance instead of Euclidean distance (ablation) | FPGM on ResNet-110 · CIFAR-10 | accuracy (%): 93.73 ± 0.23 -> 93.87 ± 0.22 | Helped |
| Use cosine distance instead of Euclidean distance (ablation) | FPGM on ResNet-110 · CIFAR-10 | accuracy (%): 93.73 ± 0.23 -> 93.56 ± 0.13 | Hurt |
| Change pruning interval from one epoch to ten epochs (ablation) | FPGM on ResNet-110 · CIFAR-10 | accuracy: less than 0.3% | No clear effect |
| Vary the amount of pruned FLOPs (ablation) | FPGM on ResNet-110 · CIFAR-10 | accuracy: 18% and 40% | Mixed |
| Add sensitivity analysis (ablation) | FPGM on VGGNet · CIFAR-10 | accuracy (%): 93.23 -> 93.54 | Helped |
Closest papers by meaning
- Pruning Filters for Efficient ConvNets (ICLR 2017 · similarity 0.90)
- Learning Filter Pruning Criteria for Deep Convolutional Neural Networks Acceleration (CVPR 2020 · similarity 0.90)
- Compressing Convolutional Neural Networks via Factorized Convolutional Filters (CVPR 2019 · similarity 0.89)
- DPFPS: Dynamic and Progressive Filter Pruning for Compressing Convolutional Neural Networks from Scratch (AAAI 2021 · similarity 0.89)
- HRank: Filter Pruning Using High-Rank Feature Map (CVPR 2020 · similarity 0.89)
- Entropy Induced Pruning Framework for Convolutional Neural Networks (AAAI 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.