Research map / Architectures and efficiency / CNN design
EffConv: Efficient Learning of Kernel Sizes for Convolution Layers of CNNs
Read the paperCheck 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 |
|---|---|---|---|
| Learn kernel sizes with EffConv instead of FlexConv | CNN kernel-size learning · CIFAR-10 | accuracy: 92.2± 0.1 -> 92.35± 0.05 | Helped |
| Learn kernel sizes with EffConv instead of FlexConv | CNN kernel-size learning · STL-10 | accuracy: 68.6 ± 0.7 -> 73.06 ± 0.53 | Helped |
| Learn kernel sizes with EffConv instead of FlexConv | CNN kernel-size learning · ImageNet-32 | Top-1 accuracy: 24.9± 0.4 -> 36.07± 0.18 | Helped |
| Increase ResNet depth while keeping the parameter budget fixed at 0.66M (ablation) | EffConv · CIFAR-10 | accuracy | Mixed |
| Increase the parameter budget for fixed-depth ResNet-20 (ablation) | EffConv · CIFAR-10 | accuracy | Helped |
| Increase the parameter budget for fixed-depth ResNet-20 (ablation) | EffConv · CIFAR-10 | training time per epoch | Mixed |
Closest papers by meaning
- Stochastic Kernel Regularisation Improves Generalisation in Deep Kernel Machines (NeurIPS 2024 · similarity 0.86)
- Fast Training of Large Kernel Models with Delayed Projections (NeurIPS 2025 · similarity 0.85)
- PeLK: Parameter-efficient Large Kernel ConvNets with Peripheral Convolution (CVPR 2024 · similarity 0.85)
- CAM Back Again: Large Kernel CNNs from a Weakly Supervised Object Localization Perspective (CVPR 2024 · similarity 0.85)
- How Feature Learning Can Improve Neural Scaling Laws (ICLR 2025 · similarity 0.85)
- Kernel Learning with Adversarial Features: Numerical Efficiency and Adaptive Regularization (NeurIPS 2025 · similarity 0.85)
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.