Research map / Label-efficient and robust learning / OOD and anomaly detection
Unified Multi-Layer Subspace Modeling for Cross-Domain OOD Detection
Gerhard Krumpl, Henning Avenhaus, Horst Possegger
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 |
|---|---|---|---|
| fuse shallow-to-deep layers then model once (PRISM) vs MDS++ | penultimate-layer Mahalanobis OOD · Average AUROC over 3 benchmarks | AUROC: 89.13 -> 93.92 | Improved |
| add residual energy term to projected Mahalanobis term (ablation) | PRISM · Average of 3 benchmarks | AUROC: 92.19 -> 93.47 | Improved |
| add residual energy term to projected Mahalanobis term (ablation) | PRISM · Average of 3 benchmarks | FPR95: 29.00 -> 23.65 | Improved |
| add projected Mahalanobis term to residual energy term (ablation) | PRISM · OpenOOD ImageNet-1k | AUROC: 76.39 -> 89.88 | Improved |
| add projected Mahalanobis term to residual energy term (ablation) | PRISM · OpenMIBOOD | AUROC: 95.72 -> 94.86 | Hurt |
| add residual energy term to projected Mahalanobis term (ablation) | PRISM · OpenOOD ImageNet-1k | AUROC: 89.94 -> 89.88 | No clear effect |
| PCA dimension 128 vs 512 (ablation) | PRISM · Average of 3 benchmarks | AUROC: 93.11 -> 93.47 | No clear effect |
| PCA dimension 768 vs 512 (ablation) | PRISM · Average of 3 benchmarks | AUROC: 93.04 -> 93.47 | No clear effect |
| multiplicative instead of additive score fusion (ablation) | PRISM · score combination | AUROC | Improved |
| OOD-calibrated layer aggregation (MDSEns) vs ID-only | MDSEns · OpenOOD ImageNet-1k | AUROC: 89.83 -> 54.41 | Hurt |
| PRISM vs MDS++ on Swin-T | PRISM · ImageNet-1k Swin-T | AUROC: 88.85 -> 90.91 | Improved |
| PRISM vs MDS++ on ViT-B/16 | PRISM · ImageNet-1k ViT-B/16 | AUROC: 89.52 -> 88.76 | Hurt |
Closest papers by meaning
- Is Complex Training Necessary for Long-Tailed OOD Detection? A Re-think from Feature Geometry (NeurIPS 2026 · similarity 0.90)
- A Systematic Analysis of Out-of-Distribution Detection Under Representation and Training Paradigm Shifts (NeurIPS 2026 · similarity 0.89)
- Mysteries of the Deep: Role of Intermediate Representations in Out of Distribution Detection (NeurIPS 2025 · similarity 0.88)
- Leveraging Perturbation Robustness to Enhance Out-of-Distribution Detection (CVPR 2025 · similarity 0.88)
- MultiOOD: Scaling Out-of-Distribution Detection for Multiple Modalities (NeurIPS 2024 · similarity 0.88)
- Enhanced OoD Detection through Cross-Modal Alignment of Multi-Modal Representations (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.
