AI Paper Atlas

Research map / Architectures and efficiency / CNN design

Designing Affine-Invariant Neural Networks for Photometric Corruption Robustness and Generalization

Mounir Messaoudi, Quentin Rapilly, Sébastien Herbreteau, Anaïs Badoual, Charles Kervrann

ICLR 2026 · Architectures and efficiency › CNN design · approach: Steerable rotation-equivariant CNNs · 0 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
use SEqSI architecture instead of StandardResNet-20 · CIFAR-10 shift corruption (no train aug)accuracy %: 51.1 -> 91.2Improved
use SEqSI architecture instead of StandardResNet-20 · CIFAR-10 affine corruption (no train aug)accuracy %: 64.1 -> 91.2Improved
use SEqSI instead of Standard on unseen spatially-varying affineResNet-20 · CIFAR-10 (no train aug)accuracy %: 31.8 -> 72.5Improved
use SEqSI instead of Standard on high noiseResNet-20 · CIFAR-10 (no train aug)accuracy %: 19.0 -> 15.2No clear effect
use SEqSI instead of Standard on clean imagesResNet-20 · CIFAR-10 original (no train aug)accuracy %: 91.7 -> 91.2Hurt
use SEqSI instead of AffEq (less restrictive constraint)ResNet-20 · CIFAR-10 original (no train aug)accuracy %: 89.6 -> 91.2Improved
add affine and non-affine training augmentation (All) to SEqSI (ablation)SEqSI ResNet-20 · CIFAR-10 spatially-varying affineaccuracy %: 72.5 -> 89.4Improved
use SEqSI instead of Standard, OOD Denoised domain3D ResNet · Cryo-ET Denoised (train WBP)accuracy %: 22.36 -> 74.53Improved
use SEqSI instead of Standard, OOD IsoNet Corrected domain3D ResNet · Cryo-ET IsoNet Corrected (train WBP)accuracy %: 15.95 -> 73.21Improved
use SEqSI instead of Standard, OOD CTF Deconvolved domain3D ResNet · Cryo-ET CTF Deconvolved (train WBP)accuracy %: 48.91 -> 66.51Improved
use SEqSI instead of Standard, in-distribution WBP3D ResNet · Cryo-ET WBP in-distributionaccuracy %: 87.17 -> 85.15Hurt
use AffEq instead of SEqSI on CTF Deconvolved3D ResNet · Cryo-ET CTF Deconvolvedaccuracy %: 66.51 -> 79.07Mixed

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