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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use magnitude-aware MLS scoring instead of MSP | CE ResNet-18 · CIFAR10 ID, OOD+OSR overall | AUROC: 91.61 -> 92.53 | Improved |
| use Energy scoring instead of MSP | CE ResNet-18 · CIFAR10 ID, OOD+OSR overall | AUROC: 91.61 -> 92.62 | Improved |
| add ReAct activation clipping to MLS | CE ResNet-18 · CIFAR10 ID, OOD+OSR overall | AUROC: 92.53 -> 91.78 | Hurt |
| add ReAct to Energy | CE ResNet-18 · CIFAR10 ID overall | AUROC: 92.62 -> 91.92 | Hurt |
| add ReAct to MLS with OE training | OE ResNet-18 · CIFAR10 ID overall | AUROC: 96.60 -> 89.34 | Hurt |
| train with Outlier Exposure instead of CE | ResNet-18 MSP · CIFAR10 ID overall | AUROC: 91.61 -> 96.44 | Improved |
| train with Outlier Exposure instead of CE at large scale | ResNet-50 MLS · ImageNet large-scale benchmark overall | AUROC: 82.71 -> 78.88 | Hurt |
| replace 300K random auxiliary images with YFCC-15M for OE (ablation) | OE ResNet-18 MLS · CIFAR10 ID overall | AUROC: 96.60 -> 93.98 | Hurt |
| replace Places with YFCC15M auxiliary data for OE (ablation) | OE ResNet-50 MLS · ImageNet large-scale benchmark overall | AUROC: 78.88 -> 71.51 | Hurt |
| add Mixup to CE training (ablation) | CE ResNet-18 MLS · CIFAR10 ID overall | AUROC: 92.53 -> 88.17 | Hurt |
| add Mixup to CE training (ablation) | CE ResNet-18 Energy · CIFAR10 ID overall | AUROC: 92.62 -> 86.32 | Hurt |
| use ARPL+CS training instead of CE | ResNet-50 MLS · ImageNet large-scale benchmark overall | AUROC: 82.71 -> 80.34 | Hurt |
Closest papers by meaning
- Expecting The Unexpected: Towards Broad Out-Of-Distribution Detection (NeurIPS 2024 · similarity 0.91)
- EAT: Towards Long-Tailed Out-of-Distribution Detection (AAAI 2024 · similarity 0.91)
- Double Descent Meets Out-of-Distribution Detection: Theoretical Insights and Empirical Analysis on the Role of Model Complexity (NeurIPS 2025 · similarity 0.91)
- ImageNet-OOD: Deciphering Modern Out-of-Distribution Detection Algorithms (ICLR 2024 · similarity 0.90)
- Test-Time Linear Out-of-Distribution Detection (CVPR 2024 · similarity 0.90)
- OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary (CVPR 2025 · similarity 0.90)
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.
