AI Paper Atlas

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

DRAUC: An Instance-wise Distributionally Robust AUC Optimization Framework

Siran Dai, Qianqian Xu, Zhiyong Yang, Xiaochun Cao, Qingming Huang

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 0 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 DRAUC-Df instead of WDROResNet20 · Tiny-ImageNet-C, DogsAUC: 82.20 -> 84.11Improved
Use DRAUC-Da instead of DRAUC-Df (ablation)DRAUC · CIFAR10-C, ResNet20, imbalance 0.05AUC: 80.18 -> 80.13Hurt
Use DRAUC-Da instead of DRAUC-Df (ablation)DRAUC · CIFAR100-LT, ResNet32, imbalance 0.05AUC: 63.90 -> 64.39Improved
Use DRAUC-Da instead of DRAUC-Df (ablation)DRAUC · CIFAR100-C, ResNet20, imbalance 0.10AUC: 66.25 -> 66.11Hurt
Use DRAUC-Da instead of DRAUC-Df (ablation)DRAUC · CIFAR10-C, ResNet32, imbalance 0.20AUC: 90.62 -> 90.15Hurt
Apply class-wise distributional perturbation rather than distribution-free perturbation (ablation)DRAUC · Corrupted benchmarks overallAUCMixed
Use DRAUC rather than AUCMLoss under matched hyperparameters (ablation)AUCMLoss · Corrupted benchmarksAUCImproved
Increase the robust diameter epsilon (ablation)DRAUC · Relatively balanced training sets, imbalance ratio at least 0.1robust performanceImproved
Increase epsilon in highly imbalanced settings (ablation)DRAUC · Highly imbalanced training setsrobust performanceNo clear effect
Use a large lambda learning rate when the imbalance ratio is 0.01 (ablation)DRAUC · Extremely imbalanced training setsrobust performanceHurt
Choose an appropriate lambda learning rate around 0.1 (ablation)DRAUC · Training sets the model can fit without difficultyrobustness: 0.1Improved
Train for robustness with DRAUC on less imbalanced dataDRAUC · Non-corrupted datasets with less extreme imbalanceAUCHurt

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