Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly 벤치마크
Unsupervised Anomaly Detection with Specified Settings -- 0.1% anomaly on Cats and Dogs
AUC-ROC
- 2008-12-15 — Isolation Forest: AUC-ROC 0.777
- 2018-01-01 — DAGMM: AUC-ROC 0.784
- 2019-03-30 — RSRAE: AUC-ROC 0.982
| Rank | Model | AUC-ROC | Extra Training Data | Paper | Code | Year |
|---|---|---|---|---|---|---|
| 1 | RSRAE | 0.982 | Robust Subspace Recovery Layer for Unsupervised Anomaly Detection | dmzou/RSRAE · marrrcin/rsrlayer-pytorch | 2019 | |
| 2 | LVAD | 0.981 | Locally varying distance transform for unsupervised visual anomaly detection | wen-yan-lin/LVAD-Locally-Varying-Anomaly-Detection | 2022 | |
| 3 | Shell-Renormalized | 0.866 | ✓ | Shell Theory: A Statistical Model of Reality | wen-yan-lin/shell-theory | 2021 |
| 4 | DAGMM | 0.784 | Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection | danieltan07/dagmm · RomainSabathe/dagmm | 2018 | |
| 5 | Isolation Forest | 0.777 | Isolation forest | 2008 | ||
| 6 | Deep Unsup. | 0.545 | Deep unsupervised anomaly detection | 2021 |