Research map / Architectures and efficiency / Pruning and efficient tuning
How Well Do Sparse ImageNet Models Transfer?
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 |
|---|---|---|---|
| Use regularization-based pruning rather than progressive sparsification for linear finetuning | ImageNet-pretrained ResNet50 · specialized downstream tasks, including Aircraft | accuracy: 40.0 $ \pm $ 0.2 -> 55.2 $ \pm $ 0.2 | Helped |
| Use progressive sparsification rather than regularization-based pruning for full finetuning | ImageNet-pretrained ResNet50 · downstream transfer tasks | accuracy | Helped |
| Increase sparsity from 90% to 98% | ImageNet-pretrained ResNet50 with full finetuning · downstream transfer tasks | accuracy | Hurt |
| Extend RigL ERK training from 1x to 5x (ablation) | ImageNet-pretrained ResNet50 · full finetuning on downstream tasks | accuracy | Helped |
| Use 90%-sparse rather than dense backbones for online linear transfer | ResNet50 with sparse inference · Caltech-101 | training time per epoch relative to dense: 1 -> $0.28\times$ | Helped |
| Transfer from a pruned rather than dense COCO model | YOLOv3 · COCO to VOC | mAP@0.5: 86.0 -> 84.0 | Hurt |
| Increase sparsity from 80% to 90% | ImageNet-pretrained ResNet50 with linear finetuning · downstream transfer tasks | accuracy | No clear effect |
| Use LTH-T at high sparsity (at least 90%) rather than other pruning methods | ImageNet-pretrained ResNet50 with full finetuning · downstream transfer tasks | accuracy | Hurt |
Closest papers by meaning
- Sparse Deep Transfer Learning for Convolutional Neural Network (AAAI 2017 · similarity 0.92)
- The Unreasonable Effectiveness of Random Pruning: Return of the Most Naive Baseline for Sparse Training (ICLR 2022 · similarity 0.88)
- Do Better ImageNet Models Transfer Better? (CVPR 2019 · similarity 0.87)
- Sparsest Models Elude Pruning: An Exposé of Pruning’s Current Capabilities (ICML 2024 · similarity 0.87)
- Sparse Model Soups: A Recipe for Improved Pruning via Model Averaging (ICLR 2024 · similarity 0.86)
- Visual Prompting Upgrades Neural Network Sparsification: A Data-Model Perspective (AAAI 2025 · 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.