atlas

Research map / Architectures and efficiency / RNNs and normalization

NAIS-Net: Stable Deep Networks from Non-Autonomous Differential Equations

NeurIPS 2018 · Architectures and efficiency › RNNs and normalization · approach: Theory of depth and width · 38 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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
unroll each NAIS-Net block 10 times instead of once (ablation)NAIS-Net · CIFAR-100 testaccuracy: 65.00 ± 0.52 -> 66.07 ± 0.24Helped
unroll each NAIS-Net block 10 times instead of once (ablation)NAIS-Net · CIFAR-10 testaccuracy: 91.24±0.10 -> 91.25±0.46No clear effect
replace ResNet with NAIS-Net (stable, non-autonomous, shared weights)ResNet · CIFAR-10generalization gap: reduced by 33%Helped
replace ResNet with NAIS-Net10ResNet · CIFAR-100 testaccuracy: 66.34 ± 0.82 -> 66.07 ± 0.24No clear effect
use NAIS-Net instead of shared-weight ResNet without BatchNorm (ablation)ResNet-SH · MNISTtest accuracy: 95.86% -> 97.28%Helped
add batch normalization to shared-weight ResNet (ablation)ResNet-SH · MNISTtest accuracy: 95.86% -> 96.69%Helped
enforce stability without non-autonomy (ResNet-SH-STABLE) (ablation)shared-weight ResNet · MNISTability to learnHurt

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