AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Limitations of Post-Hoc Feature Alignment for Robustness

Collin Burns, Jacob Steinhardt

CVPR 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Statistical alignment and hypothesis testing · 13 citations (OpenAlex)

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
Re-estimate BatchNorm statistics on target dataAdaBN on CIFAR-10-C · CIFAR-10-Caccuracy: 72.3 -> 83.6Improved
Re-estimate BatchNorm statistics on target dataAdaBN on ImageNetV2 · ImageNetV2accuracy: 63.2 -> 60.9Hurt
Re-estimate BatchNorm statistics on target dataAdaBN on Stylized ImageNet · Stylized ImageNetaccuracy: 7.1 -> 10.2Improved
Re-estimate BatchNorm statistics on target dataAdaBN on black-border shifts · CIFAR-10 and TinyImageNetaccuracy: 65.0 -> 52.5; 22.6 -> 11.8Hurt
Use training augmentations when estimating target BatchNorm statisticsAdaBN on clean CIFAR-10 and TinyImageNet · CIFAR-10 and TinyImageNetaccuracy: +2.0; +3.7Improved
Use training augmentations when estimating target BatchNorm statisticsAdaBN on ImageNet-C and Stylized ImageNet · ImageNet-C and Stylized ImageNetaccuracy: -3.6; -1.8Hurt
Use training augmentations when estimating target BatchNorm statisticsAdaBN across reported datasets · Reported datasetsaccuracy: 94.8 (+2.0); 64.0 (+3.7); 76.0 (+0.4); 86.7 (+3.1); 41.8 (+1.7); 43.3 (-3.6); 63.8 (+2.9); 8.4 (-1.8)Mixed
Exclude the final BatchNorm layers from alignment (ablation)AdaBN under CIFAR-10-C label shift · CIFAR-10-CaccuracyImproved
Exclude the first BatchNorm layers from alignment (ablation)AdaBN under CIFAR-10-C label shift · CIFAR-10-CaccuracyNo clear effect
Apply AdaBN to a smaller subset of target classes (ablation)AdaBN · CIFAR-10-Caccuracy: below 50%Hurt
Re-estimate BatchNorm statistics on target dataAdaBN · CIFAR-10-C, TinyImageNet-C, ImageNet-C, and Stylized ImageNetECE: 21.5 -> 11.3; 25.0 -> 15.2; 12.0 -> 5.2; 30.9 -> 12.9Improved
Re-estimate BatchNorm statistics on target dataAdaBN · ImageNetV2ECE: 10.6 -> 10.3Improved

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