AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

A Geometric Perspective towards Neural Calibration via Sensitivity Decomposition

Junjiao Tian, Dylan Yung, Yen-Chang Hsu, Zsolt Kira

NeurIPS 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Confidence calibration objectives · 8 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
Apply GSD with validation-set grid search of β′ResNet18 · CIFAR10 noise level 1ECE: 0.102±0.001 -> 0.040±0.002Improved
Apply GSD with validation-set grid search of β′ResNet18 · CIFAR100 noise level 5ECE: 0.237±0.01 -> 0.145±0.013Improved
Apply GSD calibrationResNet18 · CIFAR100 rotated validation setECE: 0.215±0.007 -> 0.097±0.003Improved
Apply GSD calibrationResNet18 · CIFAR10 rotated validation setNLL: 2.941±0.085 -> 2.211±0.04Improved
Disentangle the feature norm and regularize α near oneResNet18 · CIFAR10 vs. SVHNAUROC using norm: 90.48 -> 99.05Improved
Disentangle the feature norm and regularize α near oneResNet18 · CIFAR100 vs. SVHNAUROC using norm: 79.38 -> 94.46Improved
Remove input feature norm (ablation)Vanilla ResNet18 · CIFAR10, cleanECE: 0.025±0.001 -> 0.893±0.002Hurt
Remove classifier weight norm (ablation)Vanilla ResNet18 · CIFAR10, cleanECE: 0.025±0.001 -> 0.061±0.0003Hurt
Remove both classifier weight norm and input feature norm (ablation)Vanilla ResNet18 · CIFAR10, cleanECE: 0.025±0.001 -> 0.914±0.001Hurt
Remove regularization that keeps α near one (ablation)GSD ResNet18 · CIFAR100 vs. SVHNAUROC using cosine similarity: 85.68 -> 72.61Hurt

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