Research map / Label-efficient and robust learning / OOD and anomaly detection
Provable and feature-based OOD detection
Designs OOD scores and outlier training with theoretical guarantees using typical features and prototypes.
122 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2019–2026) · an approach in OOD and anomaly detection, keyword cluster “ood detection · distribution ood · distribution detection”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Few recent reports The tracked benchmarks have at most one report from 2025–26 that could beat the old record, too few to say whether they are still moving.
59 papers in 2023–24 and 39 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 27.8% to 15.1%.
Open questions the data points to
- Papers report OE on CIFAR-100 (FPR95) anywhere from 27.67 to 34.73: comparisons against it depend on whose number you take.
No longer reported by papers here: CIFAR-10 OOD average (FPR95, last in 2024); CIFAR-100 (six OOD datasets, average) (AUROC, last in 2024).
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2019 · Likelihood Ratios for Out-of-Distribution Detection (NeurIPS 2019 · 8 citations)
- 2020 · Generalized ODIN: Detecting Out-of-Distribution Image Without Learning From Out-of-Distribution Data (CVPR 2020 · 476 citations)
- 2021 · MOOD: Multi-Level Out-of-Distribution Detection (CVPR 2021 · 78 citations)
- 2022 · Zero-Shot Out-of-Distribution Detection Based on the Pre-trained Model CLIP (AAAI 2022 · 116 citations)
- 2023 · GEN: Pushing the Limits of Softmax-Based Out-of-Distribution Detection (CVPR 2023 · 85 citations)
- 2024 · ID-like Prompt Learning for Few-Shot Out-of-Distribution Detection (CVPR 2024 · 27 citations)
- 2025 · DPU: Dynamic Prototype Updating for Multimodal Out-of-Distribution Detection (CVPR 2025 · 9 citations)
- 2026 · Out-of-Distribution Detection with Positive and Negative Prompt Supervision Using Large Language Models (AAAI 2026 · 3 citations)
Earliest papers here
- Likelihood Ratios for Out-of-Distribution Detection (NeurIPS 2019 · 8 citations)
- Generalized ODIN: Detecting Out-of-Distribution Image Without Learning From Out-of-Distribution Data (CVPR 2020 · 476 citations)
- On the Value of Out-of-Distribution Testing: An Example of Goodhart's Law (NeurIPS 2020 · 86 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
CIFAR-10 · FPR95 (lower is better)
4 papers, 2023–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2023 | 11.22 | Unleashing Mask: Explore the Intrinsic Out-of-Distribution Detection Capability (ICML 2023 · 1 citation) |
CIFAR-10 OOD average · FPR95 (lower is better)
4 papers, 2022–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2022 | 23.69 | Watermarking for Out-of-distribution Detection (NeurIPS 2022 · 5 citations) |
| 2023 | 13.53 | Block Selection Method for Using Feature Norm in Out-of-Distribution Detection (CVPR 2023 · 36 citations) |
| 2024 | 11.58 | Learning to Shape In-distribution Feature Space for Out-of-distribution Detection (NeurIPS 2024 · 8 citations) |
CIFAR-100 · FPR95 (lower is better)
4 papers, 2023–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2023 | 24.8 | Diversified Outlier Exposure for Out-of-Distribution Detection via Informative Extrapolation (NeurIPS 2023 · 4 citations) |
CIFAR-100 (six OOD datasets, average) · AUROC (higher is better)
4 papers, 2023–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2023 | 92.19 | Decoupling MaxLogit for Out-of-Distribution Detection (CVPR 2023 · 47 citations) |
What has been tried
103 papers here report 1,112 measured changes (a component added, removed or swapped, compared with a baseline): helped 70%, no clear effect 6%, hurt 19%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Generate ID-like outliers from ID samples instead of using auxiliary outliers | ImageNet-1k average OOD, four-shot · AUROC | 86.15 -> 96.66 | Helped | ID-like Prompt Learning for Few-Shot Out-of-Distribution Detection (CVPR 2024 · 27 citations) |
| add tail-class prototype learning (TCPL) | CIFAR10-LT · AUC | 89.91 -> 91.23 | Helped | Out-of-Distribution Detection in Long-Tailed Recognition with Calibrated Outlier Class Learning (AAAI 2024 · 18 citations) |
| Replace selecting the most relevant dimensions with a random subspace | CIFAR-100 (six OOD datasets average) · FPR95 | 31.25 -> 42.21 | Hurt | How to Overcome Curse-of-Dimensionality for Out-of-Distribution Detection? (AAAI 2024 · 12 citations) |
| Apply RTL++ instead of RTL to MSP scores | CIFAR-100 OOD · FPR95 | 51.63 -> 43.73 | Helped | Test-Time Linear Out-of-Distribution Detection (CVPR 2024 · 12 citations) |
| Add CAP pre-rectification | CIFAR-10, six OOD datasets (average) · FPR95 | 23.65 -> 13.57 | Helped | Discriminability-Driven Channel Selection for Out-of-Distribution Detection (CVPR 2024 · 10 citations) |
| add the Gaussian mapping on top of the cosine mapping | ImageNet-1K, standard training (ResNet50), average OoD sets · FPR | 58.40 -> 48.60 | Helped | Kernel PCA for Out-of-Distribution Detection (NeurIPS 2024 · 10 citations) |
| Replace adaptive discrepancy intensification with a fixed ratio | UCF101 and Kinetics-600 near-OOD · AUROC | UCF101: 97.88 (0.1), 98.44 (0.3), 97.99 (0.5), 98.28 (0.7), DPU: 98.68; Kinetics-600: 50.05 (0.1), 50.58 (0.3), 50.10 (0 | Mixed | DPU: Dynamic Prototype Updating for Multimodal Out-of-Distribution Detection (CVPR 2025 · 9 citations) |
| Add virtual labels for OOD training samples | CIFAR10-LT, SVHN OOD · AUROC | 84.91 -> 97.92 | Helped | EAT: Towards Long-Tailed Out-of-Distribution Detection (AAAI 2024 · 9 citations) |
| Add input perturbation to ODIN | Noisy-label settings · AUROC ID vs. OOD | No clear effect | A Noisy Elephant in the Room: Is Your Out-of-Distribution Detector Robust to Label Noise? (CVPR 2024 · 9 citations) | |
| Use ℓ2-normalized rather than unnormalized class prototypes | OOD detection · OOD detection performance | Hurt | Learning to Shape In-distribution Feature Space for Out-of-distribution Detection (NeurIPS 2024 · 8 citations) | |
| add calibration from mean to BATS | ImageNet-1K ID, OOD average · FPR95 | 30.16 -> 26.73 | Helped | Exploring Channel-Aware Typical Features for Out-of-Distribution Detection (AAAI 2024 · 6 citations) |
| Add coarse-grained redundancy pruning (CRP) | CIFAR-10, six-OOD average · FPR95 | 26.55 -> 21.29 | Helped | ITP: Instance-Aware Test Pruning for Out-of-Distribution Detection (AAAI 2025 · 4 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| OE | CIFAR-100 | FPR95 | 27.67 – 34.73 | 2 papers, e.g. NeurIPS 2023 and ICML 2023 |
What reviewers will expect
Datasets and baselines named in the 27 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CIFAR-100 · 18 of 27 papers
- CIFAR-10 · 16 of 27 papers
- ImageNet-1k · 9 of 27 papers
- SVHN · 9 of 27 papers
- iNaturalist · 9 of 27 papers
- Places365 · 8 of 27 papers
- Textures · 7 of 27 papers
- MNIST · 6 of 27 papers
Baselines
- MSP · 16 of 27 papers
- Energy · 10 of 27 papers
- KNN · 10 of 27 papers
- ODIN · 9 of 27 papers
- ViM · 6 of 27 papers
- OE · 5 of 27 papers
- MaxLogit · 5 of 27 papers
- CIDER · 5 of 27 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 98 papers; from OpenAlex and the papers' affiliations).
- The University of Sydney · 7 papers
- University of Wisconsin - Madison · 7 papers
- Hong Kong Baptist University · 7 papers
- University of Technology Sydney · 6 papers
- Shanghai Jiao Tong University · 5 papers
- Cornell University · 5 papers
- Mohamed bin Zayed University of Artificial Intelligence · 4 papers
- Nanjing University of Science and Technology · 4 papers
- University of Melbourne · 4 papers
- RIKEN · 4 papers
Use it
Download BibTeX of 38 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Spurious-aware OOD detection (109 papers)
- Out-of-distribution generalization theory (71 papers)
- Uncertainty and misclassification detection (56 papers)
- Robust fine-tuning for OOD generalization (56 papers)
Elsewhere in the topic
- Unified multi-class anomaly detection (86 papers)
- Anomaly detection with contaminated data (64 papers)
- Deep one-class and tabular anomaly detection (50 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).