AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Relative stability toward diffeomorphisms indicates performance in deep nets

Leonardo Petrini, Alessandro Favero, Mario Geiger, Matthieu Wyart

NeurIPS 2021 · Learning theory and algorithms › Kernels and regression theory · approach: Generalization theory of deep networks · 6 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 R_f instead of D_f alone to predict test error (ablation)trained nets · FashionMNISTcorrelation with test error: 0.72 -> 0.94Improved
use R_f instead of D_f alone (ablation)trained nets · MNISTcorrelation with test error: 0.71 -> 0.75Improved
use R_f instead of D_f alone (ablation)trained nets · SVHNcorrelation with test error: 0.87 -> 0.81No clear effect
use R_f instead of D_f alone (ablation)trained nets · Tiny ImageNetcorrelation with test error: 0.69 -> 0.74Improved
use R_f instead of G_f alone (ablation)trained nets · FashionMNISTcorrelation with test error: -0.68 -> 0.94Improved
train EfficientNetB0 by transfer learning from ImageNet instead of from scratchEfficientNetB0 · CIFAR10test error and R_fImproved
increase training set size Ptrained nets · MNIST/FashionMNIST/CIFAR10stability to additive noise G_fHurt
increase training set size PResNet/EfficientNet · MNIST/FashionMNIST/CIFAR10stability to diffeomorphisms D_fMixed
use primitive CNN LeNet instead of SOTA netsLeNet · CIFAR10R_f: 0.6Hurt
change interpolation method for image deformations (ablation)diffeomorphism probes · MNIST/FashionMNISTR_fNo clear effect

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