Research map / Label-efficient and robust learning / OOD and anomaly detection
OoD-Bench: Quantifying and Understanding Two Dimensions of Out-of-Distribution Generalization
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 RSC instead of ERM | ResNet-18 image classification · PACS | accuracy: 81.5 $\pm$ 0.0 -> 82.8 $\pm$ 0.4 | Helped |
| Use MMD instead of ERM | ResNet-18 image classification · OfficeHome | accuracy: 63.3 $\pm$ 0.2 -> 63.8 $\pm$ 0.1 | Helped |
| Use RSC instead of ERM | ResNet-18 image classification · Terra Incognita | accuracy: 42.6 $\pm$ 0.9 -> 43.6 $\pm$ 0.5 | Helped |
| Use ERDG instead of ERM | ResNet-18 image classification · Camelyon17-WILDS | accuracy: 94.7 $\pm$ 0.1 -> 95.5 $\pm$ 0.2 | Helped |
| Use IRM instead of ERM | Two-layer perceptron classification · Colored MNIST | accuracy: 29.9 $\pm$ 0.9 -> 60.2 $\pm$ 2.4 | Helped |
| Use GroupDRO instead of ERM | ResNet-18 image classification · CelebA | accuracy: 87.2 $\pm$ 0.6 -> 87.5 $\pm$ 1.1 | No clear effect |
| Use RSC instead of ERM | ResNet-18 image classification · NICO | accuracy: 72.1 $\pm$ 1.6 -> 74.3 $\pm$ 1.9 | Helped |
| Use RSC, MMD, or SagNet rather than ERM | OoD generalization algorithms · Diversity-shift datasets versus correlation-shift datasets | ranking score: RSC: +2 -> -1; MMD: +2 -> -1; SagNet: +1 -> -2 | Mixed |
| Use VREx or GroupDRO rather than ERM | OoD generalization algorithms · Correlation-shift datasets versus diversity-shift datasets | ranking score: VREx: -1 -> +1; GroupDRO: -1 -> +1 | Mixed |
| Vary the feature extractor architecture | OoD-Bench shift estimation · Colored MNIST | estimated correlation shift: 0.8280; 0.6609; 0.6677 | No clear effect |
| Increase EfficientNet-b0 training epochs | OoD-Bench shift estimation · Colored MNIST | estimated correlation shift: 0.6465 -> 0.8099 -> 0.8428 -> 0.8211 | Mixed |
Closest papers by meaning
- OOD-Chameleon: Is Algorithm Selection for OOD Generalization Learnable? (ICML 2025 · similarity 0.89)
- Bridging OOD Detection and Generalization: A Graph-Theoretic View (NeurIPS 2024 · similarity 0.89)
- Aggregation Hides Out-of-Distribution Generalization Failures from Spurious Correlations (NeurIPS 2025 · similarity 0.89)
- Vision Transformer Neural Architecture Search for Out-of-Distribution Generalization: Benchmark and Insights (NeurIPS 2024 · similarity 0.89)
- Splitting & Integrating: Out-of-Distribution Detection via Adversarial Gradient Attribution (ICML 2025 · similarity 0.88)
- Detecting Out-of-Distribution Through the Lens of Neural Collapse (CVPR 2025 · similarity 0.88)
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.