Research map / Architectures and efficiency / Pruning and efficient tuning
EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning
Bailin Li, Bowen Wu, Jiang Su, Guangrun Wang
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 |
|---|---|---|---|
| add adaptive BN statistics recalibration before evaluating pruned candidates (ablation) | MobileNetV1 L1 pruning · CIFAR-10, FLOPs not fixed | Pearson correlation: 0.613 -> 0.933 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV1 L1 pruning · CIFAR-10, 50% FLOPs | Pearson correlation: 0.638 -> 0.854 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV1 L1 pruning · CIFAR-10, 62.5% FLOPs | Pearson correlation: 0.655 -> 0.857 | Improved |
| add adaptive BN recalibration (ablation) | ResNet-50 L1 pruning · CIFAR-10 | Pearson correlation: 0.279 -> 0.681 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV2 L1 pruning · ImageNet | Pearson correlation: -0.09 -> 0.634 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV2 L1 pruning · ImageNet | Kendall tau correlation: -0.162 -> 0.404 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV1 L1 pruning · CIFAR-10, FLOPs not fixed | Spearman correlation: 0.689 -> 0.937 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV1 L2-norm pruning · ImageNet | Pearson correlation: 0.074 -> 0.535 | Improved |
| add adaptive BN recalibration (ablation) | MobileNetV1 L2-norm pruning · ImageNet | Kendall tau correlation: 0.014 -> 0.367 | Improved |
| add adaptive BN recalibration (ablation) | ResNet-50 L1 pruning · CIFAR-10 | Spearman correlation: 0.450 -> 0.519 | Improved |
Closest papers by meaning
- Pruning from Scratch (AAAI 2020 · similarity 0.87)
- A Probabilistic Approach to Neural Network Pruning (ICML 2021 · similarity 0.85)
- “Learning-Compression” Algorithms for Neural Net Pruning (CVPR 2018 · similarity 0.85)
- On the Predictability of Pruning Across Scales (ICML 2021 · similarity 0.85)
- Pruning Neural Networks at Initialization: Why Are We Missing the Mark? (ICLR 2021 · similarity 0.84)
- Connectivity Matters: Neural Network Pruning Through the Lens of Effective Sparsity (JMLR 2023 · similarity 0.84)
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.
