AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Switchable Normalization for Learning-to-Normalize Deep Representation

Ping Luo, Ruimao Zhang, Jiamin Ren, Zhanglin Peng, Jingyu Li

TPAMI 2019 · Architectures and efficiency › RNNs and normalization · approach: Normalization layer theory · 283 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
replace BN with Switchable NormalizationResNet50 · ImageNet val, batch (8,32)top-1 accuracy (%): 76.4 -> 76.9Improved
replace BN with SN at tiny minibatchResNet50 · ImageNet val, batch (8,2)top-1 accuracy (%): 65.3 -> 75.6Improved
replace GN with SN at small minibatchResNet50 · ImageNet val, batch (8,2)top-1 accuracy (%): 75.9 -> 75.6No clear effect
replace BN with SNResNet101 · ImageNet val, batch (8,32)top-1 accuracy (%): 77.6 -> 78.6Improved
use batch average instead of moving average for BN statistics in SN (ablation)SN ResNet50 · ImageNet valtop-1 accuracy (%): 76.9 vs 76.89No clear effect
learn mixed normalizers via SN instead of IN (ablation)ResNet50 · ImageNet val, batch (8,32)top-1 accuracy (%): 71.6 -> 76.9Improved
replace LN with SN (ablation)ResNet50 · ImageNet val, batch (8,32)top-1 accuracy (%): 74.7 -> 76.9Improved
replace BRN with SN (ablation)ResNet50 · ImageNet val, batch (8,4)top-1 accuracy (%): 73.7 -> 75.9Improved
make SN sparse with argmax on control parameters (ablation)SN ResNet50 · ImageNet val, batch (8,32)top-1 accuracy (%): 76.9 -> 77.0No clear effect
finetune BN backbone in detection instead of freezing (ablation)Faster R-CNN ResNet50 · COCOAP: 29.6 -> 19.3Hurt
replace GN with SN in backbone and headFaster R-CNN+FPN · COCOAP: 38.2 -> 39.3Improved
replace GN with SN in head over frozen BN backbone (ablation)Faster R-CNN+FPN · COCOAP: 37.2 -> 38.0Improved

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