Research map / Architectures and efficiency / Pruning and efficient tuning
Towards Compact CNNs via Collaborative 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 |
|---|---|---|---|
| Jointly apply channel pruning and tensor decomposition | ResNet-50 · ImageNet 2012 | Top-1 accuracy: 75.47 -> 75.59 | Helped |
| Use collaborative compression instead of AutoML pruning | MobileNet-V2 · ImageNet 2012 | Top-1 accuracy: 70.80 -> 70.91 | Helped |
| Apply global compression rate optimization rather than a uniform per-layer rate (ablation) | CC · ResNet-50 on ImageNet 2012 | Top-1 accuracy | Helped |
| Recompute unit importance step-by-step and account for remaining compression space (ablation) | CC · ResNet-50 on ImageNet 2012 | Top-1 accuracy | Helped |
| Use a single importance calculation instead of multi-step heuristic compression (ablation) | CC · ResNet-50 on ImageNet 2012 | Top-1 accuracy | Hurt |
| Use the same compression rate for every layer instead of global rate optimization (ablation) | CC · ResNet-50 on ImageNet 2012 | Top-1 accuracy | Hurt |
| Change compression order among pruning then decomposition, decomposition then pruning, and collaborative compression (ablation) | CC with a fixed compressed structure · ResNet-56 on CIFAR-10 | Top-1 accuracy | No clear effect |
| Increase the target compression rate from C = 0.5 to C = 0.6 (ablation) | CC on ResNet-50 · ImageNet 2012 | Top-1 accuracy: 75.59 -> 74.54 | Hurt |
| Increase the target compression rate from C = 0.5 to C = 0.6 (ablation) | CC on ResNet-56 · CIFAR-10 | Top-1 accuracy: 93.87 -> 93.64 | Hurt |
| Increase the target compression rate from C = 0.4 to C = 0.6 (ablation) | CC on DenseNet-40 · CIFAR-10 | Top-1 accuracy: 94.67 -> 94.40 | Hurt |
Closest papers by meaning
- Compressing Neural Networks: Towards Determining the Optimal Layer-wise Decomposition (NeurIPS 2021 · similarity 0.91)
- Multi-Dimensional Pruning: A Unified Framework for Model Compression (CVPR 2020 · similarity 0.91)
- Distribution-Aware Tensor Decomposition for Compression of Convolutional Neural Networks (NeurIPS 2025 · similarity 0.90)
- On Compressing Deep Models by Low Rank and Sparse Decomposition (CVPR 2017 · similarity 0.89)
- OPQ: Compressing Deep Neural Networks with One-shot Pruning-Quantization (AAAI 2021 · similarity 0.89)
- PENNI: Pruned Kernel Sharing for Efficient CNN Inference (ICML 2020 · 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.