Research map / Architectures and efficiency / Pruning and efficient tuning
Automatic Network Pruning via Hilbert-Schmidt Independence Criterion Lasso under Information Bottleneck Principle
Song Guo, Lei Zhang, Xiawu Zheng, Yan Wang, Yuchao Li, Fei Chao, Chenglin Wu, Shengchuan Zhang, Rongrong Ji
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 |
|---|---|---|---|
| use Gaussian kernel instead of linear kernel in HSIC Lasso (ablation) | APIB · CIFAR-10 VGG-16, 66% FLOPs | top-1 accuracy: 92.39% -> 93.83% | Improved |
| use Gaussian kernel instead of sigmoid kernel (ablation) | APIB · CIFAR-10 VGG-16, 66% FLOPs | top-1 accuracy: 86.13% -> 93.83% | Improved |
| use Laplacian kernel instead of Gaussian kernel (ablation) | APIB · CIFAR-10 VGG-16, 66% FLOPs | top-1 accuracy: 93.83% -> 93.93% | No clear effect |
| use Gaussian kernel instead of linear kernel (ablation) | APIB · CIFAR-10 ResNet-56, 54% FLOPs | top-1 accuracy: 93.39% -> 93.92% | Improved |
| set channel threshold Omega to 5 instead of 10 at 95% sparsity (ablation) | APIB · CIFAR-10 VGG-16 | top-1 accuracy: 0.39% higher accuracy | Improved |
| set channel threshold Omega to 3 instead of 1 at 85% pruning (ablation) | APIB · CIFAR-10 ResNet-56 | top-1 accuracy: 0.3% performance improvement | Improved |
| set Omega to 0 (ablation) | APIB · CIFAR-10 | top-1 accuracy: layer collapse, model non-functional | Hurt |
| increase sample size used in pruning beyond 256 (ablation) | APIB · CIFAR-10 ResNets | selected filters and accuracy: unchanged above 256 | No clear effect |
| replace real data with random images (data-free) (ablation) | APIB · CIFAR-10 VGG-16, 60% pruned | top-1 accuracy: 94.08% -> 94.04% | No clear effect |
| replace real data with random images (data-free) (ablation) | APIB · ImageNet ResNet-50, 76% pruned | top-1 accuracy: 70.67% -> 70.76% | No clear effect |
| use APIB instead of L1 norm criterion without fine-tuning | VGG-16 · CIFAR-10 post-training, 10% sparsity | accuracy: 83.18 -> 92.44 | Improved |
| use APIB instead of Hrank without fine-tuning | VGG-16 · CIFAR-10 post-training, 30% sparsity | accuracy: 40.24 -> 91.84 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Hrank — HRank: Filter Pruning Using High-Rank Feature Map (CVPR 2020) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- An Information Theory-Inspired Strategy for Automated Network Pruning (IJCV 2025 · similarity 0.92)
- Pruning from Scratch (AAAI 2020 · similarity 0.92)
- Connectivity Matters: Neural Network Pruning Through the Lens of Effective Sparsity (JMLR 2023 · similarity 0.91)
- The Unreasonable Effectiveness of Random Pruning: Return of the Most Naive Baseline for Sparse Training (ICLR 2022 · similarity 0.90)
- Sparsest Models Elude Pruning: An Exposé of Pruning’s Current Capabilities (ICML 2024 · similarity 0.89)
- How Sparse Can We Prune A Deep Network: A Fundamental Limit Perspective (NeurIPS 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.
