Research map / Architectures and efficiency / Pruning and efficient tuning
Pruning Deep Neural Networks from a Sparsity Perspective
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 |
|---|---|---|---|
| Use SAP instead of Lottery Ticket pruning | CIFAR10 CNN and the other evaluated pruning settings · FashionMNIST and CIFAR10; additional datasets and models in the appendix | accuracy-compression trade-off | Mixed |
| Use SAP with p=1.0, q=2.0 instead of SAP with p=0.5, q=1.0 (ablation) | SAP · FashionMNIST MLP and CIFAR10 ResNet18 | compression efficiency | Mixed |
| Use neuron-wise or layer-wise pruning instead of global pruning (ablation) | SAP · CIFAR10 CNN | accuracy-compression trade-off | Helped |
| Use global pruning instead of neuron-wise or layer-wise pruning (ablation) | SAP · CIFAR10 CNN | accuracy-compression trade-off | Hurt |
| Set q=1.0 and move p closer to q (ablation) | SAP · CIFAR10 CNN, global pruning | pruning aggressiveness | Helped |
| Set p=1.0 and increase the distance between q and p (ablation) | SAP · CIFAR10 CNN, global pruning | pruning aggressiveness | Helped |
| Set η_r above zero instead of η_r=0 (ablation) | SAP · CIFAR10 CNN, global pruning | pruning aggressiveness | Helped |
| Set γ above 1 instead of γ=1 (ablation) | SAP · CIFAR10 CNN, global pruning | pruning aggressiveness | Helped |
| Use PQ Index to measure sparsity instead of Gini Index (ablation) | Pruned neural networks · Evaluated pruning procedures | sparsity measurement | No clear effect |
Closest papers by meaning
- Pruning via Sparsity-indexed ODE: a Continuous Sparsity Viewpoint (ICML 2023 · similarity 0.91)
- Computation and Memory-Efficient Model Compression with Gradient Reweighting (NeurIPS 2025 · similarity 0.91)
- Structured Pruning of Neural Networks With Budget-Aware Regularization (CVPR 2019 · similarity 0.91)
- How Sparse Can We Prune A Deep Network: A Fundamental Limit Perspective (NeurIPS 2024 · similarity 0.91)
- Targeted Low-rank Refinement: Enhancing Sparse Language Models with Precision (ICML 2025 · similarity 0.89)
- Sparsest Models Elude Pruning: An Exposé of Pruning’s Current Capabilities (ICML 2024 · 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.