Research map / Architectures and efficiency / Pruning and efficient tuning
ResRep: Lossless CNN Pruning via Decoupling Remembering and Forgetting
Xiaohan Ding, Tianxiang Hao, Jianchao Tan, Ji Liu, Jungong Han, Yuchen Guo, Guiguang Ding
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 |
|---|---|---|---|
| use 1x1 conv compactor (convolutional re-parameterization) instead of channel-wise vector (ablation) | ResRep · ImageNet-1K val (ResNet-50) | top-1 accuracy (%): 75.57 -> 76.15 | Improved |
| raise compactor momentum from 0.9 to 0.99 (ablation) | ResRep · ImageNet-1K val (ResNet-50) | FLOPs reduction (%): 45.1 -> 54.5 | Improved |
| raise compactor momentum from 0.9 to 0.99 (ablation) | ResRep · ImageNet-1K val (ResNet-50) | top-1 accuracy (%): 75.05 -> 76.15 | No clear effect |
| use ResRep instead of pruning then finetuning | ResNet-50 · ImageNet-1K val (ResNet-50) | top-1 accuracy (%): 74.66 -> 76.15 | Improved |
| use ResRep instead of a uniformly shrunk network trained from scratch | ResNet-50 · ImageNet-1K val (ResNet-50) | top-1 accuracy (%): 74.39 -> 75.97 | Improved |
| finetune the base model for 180 more epochs | ResNet-50 · ImageNet-1K val (ResNet-50) | top-1 accuracy (%): 76.15 -> 76.19 | No clear effect |
| add Rep (compactor) without Res to Lasso baseline (ablation) | Lasso pruning · CIFAR-10 ResNet-56 | accuracy at matched FLOPs reduction | Improved |
| add Res (gradient resetting) without Rep to Lasso baseline (ablation) | Lasso pruning · CIFAR-10 ResNet-56 | accuracy at matched FLOPs reduction | Improved |
| combine Rep and Res (full ResRep) (ablation) | Rep only or Res only · CIFAR-10 ResNet-56 | accuracy at matched FLOPs reduction | Improved |
| use ResRep instead of SASL at about 56% FLOPs reduction | ResNet-50 · ImageNet-1K val (ResNet-50) | top-1 accuracy drop (%): 1.00 -> 0.18 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HRank — HRank: Filter Pruning Using High-Rank Feature Map (CVPR 2020) · better on 2 of 2 shared results
- ThiNet — ThiNet: A Filter Level Pruning Method for Deep Neural Network Compression (ICCV 2017) · 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 0.90)
- Learnable Sparsity for Vision Generative Models (ICLR 2026 · similarity 0.89)
- RePr: Improved Training of Convolutional Filters (CVPR 2019 · similarity 0.89)
- Provable Filter Pruning for Efficient Neural Networks (ICLR 2020 · similarity 0.88)
- Structured Pruning of Neural Networks With Budget-Aware Regularization (CVPR 2019 · similarity 0.88)
- REPrune: Channel Pruning via Kernel Representative Selection (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.
