AI Paper Atlas

Research map / Architectures and efficiency / CNN design

1xN Pattern for Pruning Convolutional Neural Networks

Mingbao Lin, Yuxin Zhang, Yuchao Li, Bohong Chen, Fei Chao, Mengdi Wang, Shen Li, Yonghong Tian, Rongrong Ji

TPAMI 2022 · Architectures and efficiency › CNN design · approach: CNN filter pruning · 59 citations (OpenAlex)

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 changedWhereMeasuredEffect
add filter rearrangement before 1xN pruning (N=16) (ablation)1xN pruning, MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 68.971 -> 69.352Improved
add filter rearrangement before 1xN pruning (N=2) (ablation)1xN pruning, MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 69.900 -> 70.233Improved
add filter rearrangement before 1xN pruning (N=4) (ablation)1xN pruning, MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 69.521 -> 69.579Improved
add filter rearrangement before 1xN pruning (N=32) (ablation)1xN pruning, MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 68.431 -> 68.762Improved
remove blocks of consecutive kernels by l1 norm instead of random kernels (ablation)Kernel-wise pruning, ResNet-50 · ImageNet-1k val, p=50%top-1 accuracy: 74.834 -> 75.370Improved
use 1xN block-level magnitude (N=4) instead of kernel-level l1 magnitude (ablation)Kernel-wise pruning, ResNet-50 · ImageNet-1k val, p=50%top-1 accuracy: 75.370 -> 76.506Improved
use 1xN block-level magnitude (N=4) instead of kernel-level l1 magnitude (ablation)Kernel-wise pruning, MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 69.514 -> 69.706Improved
use 1x4 pattern instead of filter pruningPruned MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 66.730 -> 69.706Improved
use 1x4 pattern instead of weight pruningPruned MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 71.146 -> 69.706Hurt
increase block size N from 2 to 321xN pruning, MobileNet-V2 · ImageNet-1k val, p=50%top-1 accuracy: 70.233 -> 68.762Hurt
increase block size N from 2 to 321xN pruning, ResNet-50 · ImageNet-1k val, p=50%top-1 accuracy: 76.654 -> 75.960Hurt
use 1x4 pattern instead of weight pruning for latencyPruned MobileNet-V2 on Cortex-A7 · Inference latency, p=50%latency savings: 56.04ms inference savingsImproved

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