AI Paper Atlas

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

Neural Collapse with Normalized Features: A Geometric Analysis over the Riemannian Manifold

Can Yaras, Peng Wang, Zhihui Zhu, Laura Balzano, Qing Qu

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 4 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
Normalize features and classifiers to the sphere rather than using no normalizationUFM · UFM, 100 classes, 5 samples per classAverage accuracy: 49.9% ± 2.39% -> 100.0% ± 0.00%Improved
Normalize features and classifiers to the sphere rather than using no normalizationUFM · UFM, 100 classes, 5 samples per classAverage CE loss: 1.63 ± 0.03 -> 3.84 ± 0.00Hurt
Use feature normalization instead of regularizationResNet-18 · CIFAR100 test splitTest accuracy: 55.3% -> 58.6%Improved
Use feature normalization instead of regularizationResNet-18 · CIFAR100 test splitTest NC1: 3.838 -> 3.143Improved
Use feature normalization instead of regularizationResNet-50 · CIFAR100 test splitTest accuracy: 48.9% -> 56.4%Improved
Use feature normalization instead of regularizationResNet-50 · CIFAR100 test splitTest NC1: 4.486 -> 3.127Improved
Increase network widthResNet-18 with normalized features and classifiers · CIFAR10 with random labelsNC metricsImproved
Use feature normalization instead of regularizationResNet-18 · Reduced CIFAR100 training setTraining speed to perfect classification: 10-20 epochs soonerImproved
Use feature normalization instead of regularizationResNet-50 · Reduced CIFAR100 training setTraining accuracy: above 90%Improved
Set temperature in the range around 1 to 10, with values close to 5 (ablation)Feature-normalized ResNet-18 · CIFAR-10NC metricsImproved
Use a larger temperature outside the range that achieves perfect classification (ablation)Feature-normalized ResNet-18 · CIFAR-10Training accuracy: from about 0.1 to 10Hurt
Use label smoothing with sufficiently large temperature rather than ordinary CE (ablation)Normalized UFM · Classification under label smoothingNeural collapseHurt

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