AI Paper Atlas

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

Dissecting Out-of-Distribution Detection and Open-Set Recognition: A Critical Analysis of Methods and Benchmarks

Hongjun Wang, Sagar Vaze, Kai Han

IJCV 2024 · Label-efficient and robust learning › OOD and anomaly detection · approach: Provable and feature-based OOD detection · 25 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 magnitude-aware MLS scoring instead of MSPCE ResNet-18 · CIFAR10 ID, OOD+OSR overallAUROC: 91.61 -> 92.53Improved
use Energy scoring instead of MSPCE ResNet-18 · CIFAR10 ID, OOD+OSR overallAUROC: 91.61 -> 92.62Improved
add ReAct activation clipping to MLSCE ResNet-18 · CIFAR10 ID, OOD+OSR overallAUROC: 92.53 -> 91.78Hurt
add ReAct to EnergyCE ResNet-18 · CIFAR10 ID overallAUROC: 92.62 -> 91.92Hurt
add ReAct to MLS with OE trainingOE ResNet-18 · CIFAR10 ID overallAUROC: 96.60 -> 89.34Hurt
train with Outlier Exposure instead of CEResNet-18 MSP · CIFAR10 ID overallAUROC: 91.61 -> 96.44Improved
train with Outlier Exposure instead of CE at large scaleResNet-50 MLS · ImageNet large-scale benchmark overallAUROC: 82.71 -> 78.88Hurt
replace 300K random auxiliary images with YFCC-15M for OE (ablation)OE ResNet-18 MLS · CIFAR10 ID overallAUROC: 96.60 -> 93.98Hurt
replace Places with YFCC15M auxiliary data for OE (ablation)OE ResNet-50 MLS · ImageNet large-scale benchmark overallAUROC: 78.88 -> 71.51Hurt
add Mixup to CE training (ablation)CE ResNet-18 MLS · CIFAR10 ID overallAUROC: 92.53 -> 88.17Hurt
add Mixup to CE training (ablation)CE ResNet-18 Energy · CIFAR10 ID overallAUROC: 92.62 -> 86.32Hurt
use ARPL+CS training instead of CEResNet-50 MLS · ImageNet large-scale benchmark overallAUROC: 82.71 -> 80.34Hurt

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