Research map / Label-efficient and robust learning / OOD and anomaly detection
Deep Semi-Supervised Anomaly 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 Deep SAD instead of SSAD Raw | semi-supervised anomaly detection · CIFAR-10, labeled anomaly ratio 0.10 | AUC: 74.0$\pm$8.1 -> 79.8$\pm$7.1 | Helped |
| Use Deep SAD instead of SSAD Raw | semi-supervised anomaly detection · Fashion-MNIST, pollution ratio 0.10 | AUC: 85.6$\pm$7.0 -> 78.2$\pm$9.1 | Hurt |
| Add labeled anomalies to training | Deep SAD · CIFAR-10 | AUC: 60.9$\pm$9.4 -> 79.8$\pm$7.1 | Helped |
| Increase the number of known anomaly classes | Deep SAD · CIFAR-10 | AUC: 58.6$\pm$10.0 -> 80.4$\pm$4.6 | Helped |
| Increase unlabeled-data pollution | Deep SAD · CIFAR-10 | AUC: 77.9$\pm$7.2 -> 68.5$\pm$7.1 | Hurt |
| Vary the loss-weight hyperparameter eta (ablation) | Deep SAD · MNIST, Fashion-MNIST, and CIFAR-10 | AUC | No clear effect |
| Increase representation dimensionality (ablation) | Deep SAD · MNIST, Fashion-MNIST, and CIFAR-10 | AUC | Helped |
| Replace inverse squared norm loss with negative squared norm loss (ablation) | Deep SAD · Anomaly detection experiments | optimization stability | Hurt |
| Replace inverse squared norm loss with a hinge loss (ablation) | Deep SAD · Anomaly detection experiments | performance | Hurt |
Closest papers by meaning
- Weakly Supervised Anomaly Detection via Dual-Tailed Kernel (ICML 2025 · similarity 0.89)
- SetAD: Semi-Supervised Anomaly Learning in Contextual Sets (NeurIPS 2026 · similarity 0.88)
- Normal-Abnormal Guided Generalist Anomaly Detection (NeurIPS 2025 · similarity 0.87)
- We Need to Rethink Benchmarking in Anomaly Detection (NeurIPS 2026 · similarity 0.87)
- Contamination-Resilient Anomaly Detection via Adversarial Learning on Partially-Observed Normal and Anomalous Data (ICML 2024 · similarity 0.86)
- Anomaly Heterogeneity Learning for Open-set Supervised Anomaly Detection (CVPR 2024 · similarity 0.86)
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.