Research map / Architectures and efficiency / CNN design
Linearly Replaceable Filters for Deep Network Channel Pruning
Donggyu Joo, Eojindl Yi, Sunghyun Baek, Junmo Kim
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 |
|---|---|---|---|
| select filters by linear replaceability and compensate remaining weights when pruning | ResNet on CIFAR-10 · CIFAR-10 | accuracy: LRF-50: 92.49 -> 92.54; LRF-60: 92.49 -> 92.02 | Improved |
| select filters by linear replaceability and compensate remaining weights when pruning | ResNet-50 · ImageNet | Top-1 accuracy: 75.71 | Improved |
| add Weights Compensation to LRF rather than removing channels without changing remaining weights (ablation) | ResNet-56 · CIFAR-10, last convolution layer | output feature-map difference | Improved |
| use LRF pruning instead of training the pruned architecture from scratch (ablation) | ResNet-56 · CIFAR-10 | accuracy: 89.85 -> 93.73; 91.44 -> 93.73 | Improved |
| use LRF pruning instead of training the pruned architecture from scratch (ablation) | ResNet-56 · CIFAR-10 | accuracy: 89.81 -> 93.19; 91.13 -> 93.19 | Improved |
| vary the baseline model's accuracy while applying the same pruning method (ablation) | LRF on ResNet-56 · CIFAR-10 | pruned accuracy: 93.13 (± 0.14); 93.18 (± 0.13); 93.27 (± 0.11); 93.16 (± 0.15) | No clear effect |
| evaluate pruning by accuracy drop alone rather than final pruned accuracy | network-pruning comparisons · CIFAR-10, ResNet-56 | pruned accuracy | No clear effect |
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.
- HRank — HRank: Filter Pruning Using High-Rank Feature Map (CVPR 2020) · 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
- Reborn Filters: Pruning Convolutional Neural Networks with Limited Data (AAAI 2020 · similarity 0.91)
- Pruning Filters for Efficient ConvNets (ICLR 2017 · similarity 0.89)
- Rethinking the Pruning Criteria for Convolutional Neural Network (NeurIPS 2021 · similarity 0.89)
- Convolutional Neural Network Pruning With Structural Redundancy Reduction (CVPR 2021 · similarity 0.89)
- Gate Decorator: Global Filter Pruning Method for Accelerating Deep Convolutional Neural Networks (NeurIPS 2019 · similarity 0.88)
- 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.
