AI Paper Atlas

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

Is Importance Weighting Incompatible with Interpolating Classifiers?

Ke Alexander Wang, Niladri Shekhar Chatterji, Saminul Haque, Tatsunori Hashimoto

ICLR 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 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
Replace reweighted cross-entropy with reweighted polynomial-tailed lossInterpolating neural networks · Imbalanced binary CIFAR10 and subsampled CelebAtest accuracy: as much as 9\%Improved
Exponentiate the importance weights (ablation)Polynomial-tailed loss · Imbalanced binary CIFAR10 and subsampled CelebAtest accuracyImproved
Exponentiate the importance weights (ablation)Cross-entropy loss · Imbalanced binary CIFAR10 and subsampled CelebAtest accuracyNo clear effect
Use importance weights rather than no importance weighting (ablation)Polynomial-tailed loss · Imbalanced binary CIFAR10 and subsampled CelebAtest accuracyImproved
Use early stopping (ablation)Importance-weighted cross-entropy and polynomial-tailed loss · Imbalanced binary CIFAR10 and subsampled CelebAtest accuracyImproved
Use polynomial-tailed loss rather than cross-entropy with early stopping (ablation)Importance-weighted neural networks · Imbalanced binary CIFAR10test accuracyImproved
Use minority-class weight w scaling with τ³ rather than unbiased weight τ (ablation)Polynomial-tailed linear classifier · Label shift, sub-Gaussian clusterstest error: τ³Improved
Use importance weights with polynomially-tailed loss rather than exponentially-tailed lossOverparameterized linear classifier · Label shift, sub-Gaussian clusterstest error: 1/8Improved
Use polynomial-tailed loss with DRO rather than VS loss with DROCelebA classifiers · Subsampled CelebAtest accuracy: 3.5\%Improved
Use polynomial-tailed loss with DRO rather than VS loss with DROCelebA classifiers · Subsampled CelebAworst group accuracy: 8\%Improved

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