Research map / Label-efficient and robust learning / OOD and anomaly detection
Block Selection Method for Using Feature Norm in Out-of-Distribution Detection
Read the paperCheck 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 selected-block FeatureNorm instead of output-based OOD scores | ResNet18 on CIFAR10 · CIFAR10 OOD average | FPR95: 29.14 -> 25.67 | Helped |
| Use selected-block FeatureNorm instead of output-based OOD scores | WRN28 on CIFAR10 · CIFAR10 OOD average | FPR95: 28.65 -> 13.53 | Helped |
| Use selected-block FeatureNorm instead of output-based OOD scores | VGG11 on CIFAR10 · CIFAR10 OOD average | FPR95: 46.46 -> 39.34 | Helped |
| Use selected-block FeatureNorm instead of the strongest baseline | ResNet50 on ImageNet · ImageNet OOD average | FPR95: 31.43 -> 35.45 | Hurt |
| Use selected-block FeatureNorm instead of output-based OOD scores | VGG16 on ImageNet · ImageNet OOD average | FPR95: 53.53 -> 27.55 | Helped |
| Use selected-block FeatureNorm instead of output-based OOD scores | MobileNetV3 on ImageNet · ImageNet OOD average | FPR95: 49.39 -> 35.64 | Helped |
| Replace the last-block norm with the selected-block norm (ablation) | MSP on ResNet18 · CIFAR10 OOD average | FPR95: 50.63 -> 35.56 | Helped |
| Replace the last-block norm with the selected-block norm (ablation) | ODIN on ResNet18 · CIFAR10 OOD average | FPR95: 29.14 -> 15.99 | Helped |
| Replace the last-block norm with the selected-block norm (ablation) | Energy on ResNet18 · CIFAR10 OOD average | FPR95: 29.66 -> 17.59 | Helped |
| Use BN-ReLU-Conv instead of Conv-BN-ReLU block order (ablation) | ResNet18 · CIFAR10 OOD average | FPR95: 25.67 -> 21.03 | Helped |
Closest papers by meaning
- Fast Decision Boundary based Out-of-Distribution Detector (ICML 2024 · similarity 0.88)
- Image Background Serves as Good Proxy for Out-of-distribution Data (ICLR 2024 · similarity 0.88)
- Enhancing the Power of OOD Detection via Sample-Aware Model Selection (CVPR 2024 · similarity 0.88)
- Detecting Out-of-Distribution Through the Lens of Neural Collapse (CVPR 2025 · similarity 0.87)
- Exploiting Discrepancy in Feature Statistic for Out-of-Distribution Detection (AAAI 2024 · similarity 0.87)
- Is Complex Training Necessary for Long-Tailed OOD Detection? A Re-think from Feature Geometry (NeurIPS 2026 · similarity 0.87)
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.