AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

ODP-Bench: Benchmarking Out-of-Distribution Performance Prediction

Han Yu, Kehan Li, Dongbai Li, Yue He, Xingxuan Zhang, Peng Cui

ICCV 2025 · Label-efficient and robust learning › OOD and anomaly detection · approach: Provable and feature-based OOD detection · 1 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
train models with RSC instead of ERM (ablation)performance prediction algorithms · PACSaverage Spearman rho: 0.553 -> 0.595Improved
train models with Mixup instead of ERM (ablation)performance prediction algorithms · PACSaverage Spearman rho: 0.553 -> 0.620Improved
train models with SWAD instead of ERM (ablation)performance prediction algorithms · PACSaverage Spearman rho: 0.553 -> 0.569Improved
train models with CORAL instead of ERM (ablation)performance prediction algorithms · PACSaverage Spearman rho: 0.553 -> 0.569Improved
train models with RSC instead of ERM (ablation)MaNo · PACSSpearman rho: 0.489 -> 0.788Improved
train models with Mixup instead of ERM (ablation)COTT · PACSSpearman rho: 0.636 -> 0.831Improved
use MoCo/CLIP instead of supervised pretraining (ablation)ATC, NI, COTT · PACSSpearman rhoImproved
use supervised instead of MoCo/CLIP pretraining (ablation)Agreement · PACSSpearman rhoImproved
switch from ResNet-50 to larger ViT-B/16 (ablation)performance prediction algorithms (except ATC) · PACSSpearman rhoHurt
increase corruption severity (ablation)NI · ImageNet-C and TinyImageNet-CSpearman rhoHurt
evaluate with rho@10% instead of rho (ablation)MaNo · ImageNet-Aranking metric: 0.417 -> 0.927Improved
evaluate on natural shift (CelebA) instead of synthetic corruption (CIFAR-10-C)average of 10 algorithms · OOD datasetsSpearman rho: 0.746 -> 0.042Hurt

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