Research map / Probabilistic and scientific modeling / Bayesian optimization
Bayesian Optimization with Clustering and Rollback for CNN Auto Pruning
Hanwei Fan, Jiandong Mu, Wei Zhang
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 layer clustering | Naive BO ResNet56 · CIFAR-10 | top-1 accuracy (%): 90.29 -> 91.68 | Improved |
| add rollback after layer clustering | Layer Clustering ResNet56 · CIFAR-10 | top-1 accuracy (%): 91.68 -> 92.53 | Improved |
| add layer clustering | Naive BO MobileNetV1 · ImageNet | top-1 accuracy (%): 47.39 -> 48.49 | Improved |
| add rollback after layer clustering | Layer Clustering MobileNetV1 · ImageNet | top-1 accuracy (%): 48.49 -> 49.34 | Improved |
| add layer clustering | Naive BO MobileNetV2 · ImageNet | top-1 accuracy (%): 51.09 -> 52.57 | Improved |
| add direct rollback after layer clustering | Layer Clustering MobileNetV2 · ImageNet | top-1 accuracy (%): 52.57 -> 52.86 | Improved |
| use gradual rollback instead of direct rollback (ablation) | Layer Clustering (Structure) MobileNetV2 · ImageNet | top-1 accuracy gain (points): +0.41 -> +1.60 | Improved |
| use gradual rollback instead of direct rollback (ablation) | Layer Clustering (Euclidean) MobileNetV2 · ImageNet | top-1 accuracy gain (points): +0.29 -> +0.44 | Improved |
| replace RL agent with Bayesian optimization | RL pruning agent MobileNetV2 · ImageNet | top-1 accuracy (%): 43.15 -> 51.09 | Improved |
| use structure similarity instead of Euclidean distribution similarity (ablation) | Layer Clustering ResNet56 · CIFAR-10 | top-1 accuracy (%): 88.95 -> 91.68 | Improved |
| use Euclidean distribution similarity instead of structure similarity (ablation) | Layer Clustering MobileNetV2 · ImageNet | top-1 accuracy (%): 50.31 -> 52.57 | Improved |
| use Euclidean instead of JS divergence similarity (ablation) | Layer Clustering MobileNetV1 · ImageNet | top-1 accuracy (%): 47.99 -> 48.50 | Improved |
Closest papers by meaning
- Accelerate CNN via Recursive Bayesian Pruning (ICCV 2019 · similarity 0.89)
- Multi-Dimensional Pruning: A Unified Framework for Model Compression (CVPR 2020 · similarity 0.87)
- Auto-Balanced Filter Pruning for Efficient Convolutional Neural Networks (AAAI 2018 · similarity 0.87)
- SPDY: Accurate Pruning with Speedup Guarantees (ICML 2022 · similarity 0.87)
- BMRS: Bayesian Model Reduction for Structured Pruning (NeurIPS 2024 · similarity 0.87)
- Lightweight Dataset Pruning without Full Training via Example Difficulty and Prediction Uncertainty (ICML 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.
