AI Paper Atlas

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

Diagnosing Pretrained Models for Out-of-distribution Detection

Haipeng Xiong, Kai Xu, Angela Yao

ICCV 2025 · Label-efficient and robust learning › OOD and anomaly detection · approach: Provable and feature-based OOD detection · 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
train with the v2 recipe (label smoothing, mixup, etc.) instead of v1ResNet50 · ImageNet-1kOOD AUROC: 4% ... 12%Hurt
add mixup to the v1 recipe (ablation)ResNet18 v1 · ImageNet200AUROC: 87.00 -> 84.00Hurt
replace mixup with regmixup (ablation)ResNet18 v1 · ImageNet200AUROC: 87.00 -> 86.97No clear effect
use mixup-AugRevise with virtual separation loss (ablation)ResNet18 v1 · ImageNet200AUROC: 87.00 -> 87.72Improved
apply AugDelete (refit last layer) to v2* (ablation)ResNet18 v2* · ImageNet200AUROC: 82.78 -> 85.57Improved
train with AugRevise instead of v2* recipeResNet18 · ImageNet200AUROC: 82.78 -> 87.88Improved
add label smoothing to AugRevise (ablation)AugRevise · ImageNet200AUROC: 87.88 -> 87.17Hurt
apply AugDelete with MLS scoreResNet50-v2 · ImageNet-1kAUROC: 72.84 -> 83.08Improved
apply AugDelete with ASH scoreResNet50-v2 · ImageNet-1kAUROC: 53.53 -> 81.70Improved
apply AugDelete with KNN scoreResNet50-v2 · ImageNet-1kAUROC: 79.91 -> 79.91No clear effect
apply AugDelete with NNGuide scoreResNet50-v2 · ImageNet-1kAUROC: 65.77 -> 77.54Improved
use AugRevise training with NNGuide score instead of v1ResNet18 · ImageNet200AUROC: 87.83 -> 89.31Improved

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