Research map / Label-efficient and robust learning / OOD and anomaly detection
Mitigating Simplicity Bias in OOD Detection through Object Co-occurrence Analysis
Boyang Dai, Chaoqi Chen, Yizhou Yu
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 |
|---|---|---|---|
| add reconstruction loss L_aux (ablation) | OCO · ImageNet-200 OOD | AUROC: 92.77 -> 93.69 | Improved |
| add OCO score without L_aux (ablation) | slot-based classifier · ImageNet-200 OOD | AUROC: 90.76 -> 89.43 | Hurt |
| add L_aux when no OCO score used (ablation) | slot-based classifier · ImageNet-200 OOD | AUROC: 90.76 -> 92.77 | Improved |
| add OCO score given L_aux (ablation) | slot-based classifier with L_aux · ImageNet-200 OOD | FPR95: 30.70 -> 28.23 | Improved |
| add OCO score given L_aux (ablation) | slot-based classifier with L_aux · ImageNet-200 FS-OOD | AUROC: 67.74 -> 69.56 | Improved |
| add OCO score without L_aux (ablation) | slot-based classifier · ImageNet-200 FS-OOD | FPR95: 72.50 -> 74.70 | Hurt |
| use scenario-specific scores (S1 single pattern) (ablation) | OCO · ImageNet-200 OOD, scenario S1 | FPR95: FPR95 decreased by 17.28% | Improved |
| use scenario-specific scores (S2 typical pattern) (ablation) | OCO · ImageNet-200 OOD, scenario S2 | AUROC: AUROC increased by 7.41%; FPR95 decreased by 16.79% | Improved |
| use scenario-specific scores (S3 atypical pattern) (ablation) | OCO · ImageNet-200 OOD, scenario S3 | AUROC: AUROC increased by 8.13%; FPR95 decreased by 11.53% | Improved |
| increase slot count up to K=6 (ablation) | OCO · ImageNet-200 OOD | FPR95/AUROC | Improved |
| increase slot count beyond K=6 (ablation) | OCO · ImageNet-200 OOD | FPR95/AUROC | Hurt |
| use object-centric classification (label in any slot prediction) | slot classifier · ImageNet-200 | accuracy: 94.04 -> 96.51 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- OODD — OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary (CVPR 2025) · better on 6 of 7 shared results
- NECO — NECO: NEural Collapse Based Out-of-distribution detection (ICLR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning Latent Concepts for Detecting Out-of-Distribution Objects (CVPR 2026 · similarity 0.91)
- Image Background Serves as Good Proxy for Out-of-distribution Data (ICLR 2024 · similarity 0.90)
- RUNA: Object-Level Out-of-Distribution Detection via Regional Uncertainty Alignment of Multimodal Representations (AAAI 2025 · similarity 0.89)
- Is Complex Training Necessary for Long-Tailed OOD Detection? A Re-think from Feature Geometry (NeurIPS 2026 · similarity 0.88)
- A Systematic Analysis of Out-of-Distribution Detection Under Representation and Training Paradigm Shifts (NeurIPS 2026 · similarity 0.88)
- Conjugated Semantic Pool Improves OOD Detection with Pre-trained Vision-Language Models (NeurIPS 2024 · 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.
