AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

CrossNorm and SelfNorm for Generalization Under Distribution Shifts

Zhiqiang Tang, Yunhe Gao, Yi Zhu, Zhi Zhang, Mu Li, Dimitris N. Metaxas

ICCV 2021 · Label-efficient and robust learning › OOD and anomaly detection · approach: Robust fine-tuning for OOD generalization · 85 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 CrossNorm and SelfNorm togetherWideResNet-40-2 etc. · CIFAR-100-C (mean over 4 backbones)mCE (%): 55.6 -> 43.5Improved
add CrossNorm and SelfNorm on top of AugMixAugMix · CIFAR-10-C (mean over 4 backbones)mCE (%): 12.5 -> 10.3Improved
add CNSN and AugMix (with IBN) on ImageNetResNet50 AugMix · ImageNet-CmCE (%): 68.4 -> 62.8Improved
add CrossNorm to FixMatch with RandAugment, 250 labelsFixMatch WideResNet-28-2 · CIFAR-10-C semi-supervised (250 labels, RandAugment)mCE (%): 23.4 -> 16.8Improved
add CrossNorm to FixMatch with weak augmentation, 250 labelsFixMatch WideResNet-28-2 · CIFAR-10 semi-supervised (250 labels, weak aug)mCE (%): 62.9 -> 56.5Improved
add CrossNorm and SelfNorm to FCN for domain generalizationFCN ResNet50 · GTA5->Cityscapes (target)mIoU: 21.4 -> 36.5Improved
add CrossNorm to FCN, measured on source domain versus domain randomizationFCN ResNet50 · GTA5->Cityscapes (source GTA5)mIoU: 49.0 -> 61.2Improved
add CrossNorm and SelfNorm to ConvNet sentiment classifierGloVe ConvNet · IMDb->SST-2 (target SST-2)accuracy (%): 71.9 -> 74.9Improved
add CrossNorm to ResNet50 for unsupervised domain adaptationResNet50 · Office-31 domain adaptation (avg)accuracy (%): 76.1 -> 79.2Improved
use 2-instance instead of 1-instance CrossNorm (ablation)CN WideResNet-40-2 · CIFAR-100-C (WideResNet-40-2)mCE (%): 50.5 -> 48.8Improved
add style cropping to 2-instance CrossNorm (ablation)CN WideResNet-40-2 · CIFAR-100-C (WideResNet-40-2)mCE (%): 48.8 -> 47.9Improved
use SelfNorm instead of SE attention (ablation)WideResNet-40-2 · CIFAR-100-C (WideResNet-40-2)mCE (%): 52.3 -> 47.4Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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