AI Paper Atlas

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

Inducing Neural Collapse in Imbalanced Learning: Do We Really Need a Learnable Classifier at the End of Deep Neural Network?

Yibo Yang, Shixiang Chen, Xiangtai Li, Liang Xie, Zhouchen Lin, Dacheng Tao

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Loss design and classifier theory · 21 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
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossResNet · CIFAR-10, τ=0.005accuracy (%): 67.3 -> 71.9Improved
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossResNet · CIFAR-100, τ=0.005accuracy (%): 38.7 -> 40.9Improved
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossResNet · SVHN, τ=0.01accuracy (%): 40.9 -> 45.7Improved
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossResNet · STL-10, τ=0.01accuracy (%): 37.9 -> 37.2Hurt
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossDenseNet · CIFAR-100, τ=0.005accuracy (%): 40.3 -> 40.1Hurt
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossResNet-50 · ImageNet-LT, 90 epochsaccuracy (%): 34.6 -> 41.8Improved
Replace the learnable classifier and CE with a fixed ETF classifier and DR lossResNet-34 · CUB-200-2011accuracy (%): 82.2 -> 83.0Improved
Replace the learnable classifier with a fixed ETF classifier while retaining CE (ablation)ResNet · CIFAR-10, τ=0.005, without Mixupaccuracy (%): 66.1 -> 60.4Hurt
Replace CE with DR while using the fixed ETF classifier (ablation)ResNet · CIFAR-10, τ=0.005, without Mixupaccuracy (%): 60.4 -> 68.4Improved
Add class-distribution weighting to CE (ablation)Learnable classifier + CE · CIFAR-10, τ=0.005, without Mixupaccuracy (%): 66.1 -> 66.8Improved
Use Mixup with ETF classifier + DR (ablation)ResNet · CIFAR-10, τ=0.005accuracy (%): 68.4 -> 71.9Improved
Use Mixup with ETF classifier + CE (ablation)ResNet · CIFAR-10, τ=0.01accuracy (%): 72.9 -> 67.0Hurt

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