Research map / Architectures and efficiency / CNN design
Exploiting Kernel Sparsity and Entropy for Interpretable CNN Compression
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 KSE-guided kernel clustering with G=4 | ResNet-56 · CIFAR-10 | Top-1 accuracy: 93.03 -> 93.23 | Helped |
| Use KSE-guided kernel clustering with G=5 | ResNet-56 · CIFAR-10 | Top-1 accuracy: 93.03 -> 92.88 | Hurt |
| Use KSE-guided kernel clustering | DenseNet-40 · CIFAR-10 | Top-1 accuracy: 94.81 -> 94.81 | No clear effect |
| Use KSE-guided kernel clustering | DenseNet-BC-100 · CIFAR-10 | Top-1 accuracy: 95.45 -> 95.49 | Helped |
| Use KSE-guided kernel clustering | DenseNet-40-12 · SVHN | Top-1 accuracy: 98.19 -> 98.25 | Helped |
| Use KSE-guided kernel clustering | ResNet-50 · ImageNet 2012 | Top-5 accuracy: 90.02 -> 92.52 | Helped |
| Increase compression granularity from G=2 to G=4 (ablation) | ResNet-56 with KSE · CIFAR-10 | Top-1 accuracy: 0.52% | Helped |
| Use fine-grained kernel allocation rather than coarse keep-or-prune allocation (ablation) | KSE · CIFAR-10 | Top-1 accuracy: 0.52% | Helped |
| Use the combined KSE indicator instead of kernel sparsity alone (ablation) | ResNet-56 with kernel clustering · CIFAR-10 | Top-1 accuracy | Helped |
| Use the combined KSE indicator instead of kernel entropy alone (ablation) | ResNet-56 with kernel clustering · CIFAR-10 | Top-1 accuracy | Helped |
Closest papers by meaning
- PENNI: Pruned Kernel Sharing for Efficient CNN Inference (ICML 2020 · similarity 0.88)
- Towards Compact CNNs via Collaborative Compression (CVPR 2021 · similarity 0.88)
- On Compressing Deep Models by Low Rank and Sparse Decomposition (CVPR 2017 · similarity 0.87)
- Compressing Neural Networks: Towards Determining the Optimal Layer-wise Decomposition (NeurIPS 2021 · similarity 0.87)
- Distribution-Aware Tensor Decomposition for Compression of Convolutional Neural Networks (NeurIPS 2025 · similarity 0.86)
- Convolutional Neural Network Compression through Generalized Kronecker Product Decomposition (AAAI 2022 · 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.