Research map / Architectures and efficiency / Pruning and efficient tuning
DepGraph: Towards Any Structural Pruning
Read the paperCheck 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 |
|---|---|---|---|
| add sparse learning (ablation) | DepGraph pruning · CIFAR-10, ResNet-56 | accuracy: 93.46 -> 93.64 | Helped |
| add sparse learning (ablation) | DepGraph pruning · CIFAR-100, VGG-19 | accuracy: 67.60 -> 70.39 | Helped |
| use full grouping instead of no grouping (ablation) | DepGraph pruning · CIFAR-100, ResNet-56, 3.0× learned sparsity | accuracy: 67.89 -> 70.38 | Helped |
| use full grouping instead of convolution-only grouping (ablation) | DepGraph pruning · CIFAR-100, ResNet-56, 3.0× learned sparsity | accuracy: 69.07 -> 70.38 | Helped |
| use full grouping instead of no grouping (ablation) | DepGraph pruning · CIFAR-100, VGG-19, 3.0× learned sparsity | accuracy: 53.21 -> 72.46 | Helped |
| use full grouping instead of convolution-only grouping (ablation) | DepGraph pruning · CIFAR-100, VGG-19, 3.0× learned sparsity | accuracy: 71.80 -> 72.46 | Helped |
| use full grouping instead of no grouping (ablation) | DepGraph pruning · CIFAR-100, DenseNet-121, 3.0× learned sparsity | accuracy: 78.62 -> 79.19 | Helped |
| use learned sparsity instead of uniform sparsity (ablation) | Full Grouping pruning · CIFAR-100, ResNet-56 | accuracy: 68.70 -> 70.38 | Helped |
| use learned sparsity instead of uniform sparsity (ablation) | Full Grouping pruning · CIFAR-100, VGG-19 | accuracy: 71.57 -> 72.46 | Helped |
| use learned sparsity instead of uniform sparsity (ablation) | Full Grouping pruning · CIFAR-100, DenseNet-121 | accuracy: 77.97 -> 79.19 | Mixed |
| use learned sparsity instead of uniform sparsity (ablation) | Full Grouping pruning · CIFAR-100, MobileNetV2 | accuracy: 70.06 -> 70.81 | Helped |
| use learned sparsity instead of uniform sparsity (ablation) | Full Grouping pruning · CIFAR-100, GoogleNet | accuracy: 76.90 -> 77.00 | Mixed |
Closest papers by meaning
- MDP: Multidimensional Vision Model Pruning with Latency Constraint (CVPR 2025 · similarity 0.88)
- GDeR: Safeguarding Efficiency, Balancing, and Robustness via Prototypical Graph Pruning (NeurIPS 2024 · similarity 0.88)
- Sparsest Models Elude Pruning: An Exposé of Pruning’s Current Capabilities (ICML 2024 · similarity 0.86)
- DEPrune: Depth-wise Separable Convolution Pruning for Maximizing GPU Parallelism (NeurIPS 2024 · similarity 0.86)
- Automatic Joint Structured Pruning and Quantization for Efficient Neural Network Training and Compression (CVPR 2025 · similarity 0.86)
- UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer (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.