AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Filter Grafting for Deep Neural Networks

Fanxu Meng, Hao Cheng, Ke Li, Zhixin Xu, Rongrong Ji, Xing Sun, Guangming Lu

CVPR 2020 · Architectures and efficiency › CNN design · approach: CNN filter pruning · 35 citations (OpenAlex)

Check 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 changedWhereMeasuredEffect
Graft external network weights into MobileNetV2MobileNetV2 · CIFAR-100accuracy: 71.44 -> 72.90Improved
Use external scions rather than internal filters (ablation)MobileNetV2 filter grafting · CIFAR-10accuracy: 92.68 -> 92.94Improved
Use entropy rather than L1 norm to measure layer information (ablation)ResNet32 grafting · CIFAR-10accuracy: 93.24 -> 93.33Improved
Use entropy rather than L1 norm to measure layer information (ablation)MobileNetV2 grafting · CIFAR-100accuracy: 72.90 -> 73.26Improved
Vary sampling order across grafted models (ablation)MobileNetV2 grafting · CIFAR-100accuracy: 71.91 -> 73.26Improved
Vary learning rates as well as sampling order (ablation)MobileNetV2 grafting · CIFAR-100accuracy: 73.26 -> 74.15Improved
Increase the number of grafted networks from 6 to 8 (ablation)MobileNetV2 grafting · CIFAR-100accuracy: 78.32 -> 77.76Hurt
Graft complete layer weights rather than selected filters (ablation)ResNet32 grafting · CIFAR-100accuracy: 70.79 -> 71.28Improved
Calculate entropy over the whole layer rather than summing filter entropies (ablation)ResNet32 grafting · CIFAR-100accuracy: 70.95 -> 71.28Improved
Use adaptive weighting rather than a fixed coefficient of 0.5 (ablation)ResNet32 grafting · CIFAR-100accuracy: 70.70 -> 71.28Improved
Graft weights using six networks rather than one baseline networkMobileNetV2 · CIFAR-100accuracy: 71.44 -> 78.32Improved
Graft weights between networksResNet18 · ImageNettop-1: 69.15 -> 71.19Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy