Research map / Detection and segmentation / CNN features
Differentiable Kernel Evolution
Yu Liu, Jihao Liu, Ailing Zeng, Xiaogang 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 |
|---|---|---|---|
| replace conv layers with DKE | AlexNet · ImageNet | top-1 error (%): 42.8 -> 39.1 | Improved |
| replace conv layers with DKE | ResNet-18 · ImageNet | top-1 error (%): 30.8 -> 27.15 | Improved |
| replace conv layers with DKE | ResNet-101 · ImageNet | top-1 error (%): 22.7 -> 21.1 | Improved |
| replace conv layers with DKE | AlexNet · CIFAR-100 | top-1 error (%): 56.30 -> 53.1 | Improved |
| replace conv layers with DKE | ResNet-101 · CIFAR-10 | top-1 error (%): 5.3 -> 4.7 | Improved |
| replace DCN by DKE layers | ResNet-18 · MegaFace 1 vs 1 million | top-1 accuracy (%): 69.5 -> 91.5 | Improved |
| replace conv layers with DKE | ResNet-18 · MegaFace 1 vs 1 million | top-1 accuracy (%): 89.4 -> 91.5 | Improved |
| replace conv layers with DKE | ResNet-101 · MegaFace 1 vs 1 million | top-1 accuracy (%): 97.8 -> 98.0 | Improved |
| replace conv layers with deformable convolution (DCN) | ResNet-18 · MegaFace 1 vs 1 million | top-1 accuracy (%): 89.4 -> 69.5 | Hurt |
| replace conv layers with DKE | Faster-RCNN Res18 · COCO 2017 miniVal | mAP[.5:.95]: 28.0 -> 29.6 | Improved |
| replace conv layers with DKE | FPN Res50 · COCO 2017 miniVal | mAP[.5:.95]: 36.3 -> 36.7 | Improved |
| enlarge kernel to 5x5 instead of 3x3 (ablation) | ResNet-6 · ImageNet | top-1 error (%): 51.7 -> 46.6 | Improved |
Closest papers by meaning
- Progressive Differentiable Architecture Search: Bridging the Depth Gap Between Search and Evaluation (ICCV 2019 · similarity 0.84)
- Towards Improving the Consistency, Efficiency, and Flexibility of Differentiable Neural Architecture Search (CVPR 2021 · similarity 0.84)
- DrNAS: Dirichlet Neural Architecture Search (ICLR 2021 · similarity 0.83)
- Differentiable Model Scaling using Differentiable Topk (ICML 2024 · similarity 0.83)
- Neural Fine-Tuning Search for Few-Shot Learning (ICLR 2024 · similarity 0.83)
- Adaptive Kernel Design for Bayesian Optimization Is a Piece of CAKE with LLMs (NeurIPS 2025 · similarity 0.82)
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.
