Research map / Time series / Anomaly detection
Complexity- and Statistics-Guided Anomaly Detection in Time Series Foundation Models
Jongwon Kim, Samuel Yoon, Young Myoung Ko, Yerin Kim, Sung Il Kim, JAEUNG TAE
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 |
|---|---|---|---|
| re-inject instance mean and standard deviation before the linear decoder (MOMENT-Stat) (ablation) | MOMENT · TSB-AD-U | VUS-PR: 0.3857 -> 0.3913 | Improved |
| MOMENT-Stat versus zero-shot MOMENT | MOMENT · TSB-AD-U | VUS-PR: 0.3790 -> 0.3913 | Improved |
| add complexity-aware ensemble with MOMENT-Stat | Sub-PCA · TSB-AD-U | VUS-PR: 0.4233 -> 0.4679 | Improved |
| add complexity-aware ensemble with MOMENT-Stat | Sub-IForest · TSB-AD-U | VUS-PR: 0.2230 -> 0.4318 | Improved |
| add complexity-aware ensemble with MOMENT-Stat | SR · TSB-AD-U | VUS-PR: 0.3237 -> 0.4596 | Improved |
| add complexity-aware ensemble with MOMENT-Stat | Sub-HBOS · TSB-AD-U | VUS-PR: 0.2283 -> 0.3734 | Mixed |
| use continuous alpha weighting (CAE) instead of random selection (ablation) | Sub-PCA + MOMENT-Stat · TSB-AD-U | VUS-PR: 0.4073 -> 0.4679 | Improved |
| use continuous alpha weighting (CAE) instead of hard selection (CAS) (ablation) | Sub-PCA + MOMENT-Stat · TSB-AD-U | VUS-PR: 0.4400 -> 0.4679 | Improved |
| use dataset-level alpha instead of instance-level alpha (ablation) | CAE Sub-PCA · TSB-AD-U | VUS-PR: 0.4582 -> 0.4679 | Improved |
| use proposed complexity measure instead of Spectral Entropy (ablation) | CAE Sub-PCA · TSB-AD-U | VUS-PR: 0.4288 -> 0.4679 | Improved |
| use proposed complexity measure instead of Spectral Entropy (ablation) | CAE Sub-HBOS · TSB-AD-U | VUS-PR: 0.3835 -> 0.3734 | Hurt |
| ensemble two statistical models instead of statistical plus MOMENT-Stat (ablation) | best Stat+Stat pair · TSB-AD-U | VUS-PR: 0.4398 -> 0.4679 | Hurt |
Closest papers by meaning
- The Elephant in the Room: Towards A Reliable Time-Series Anomaly Detection Benchmark (NeurIPS 2024 · similarity 0.90)
- Can LLMs Understand Time Series Anomalies? (ICLR 2025 · similarity 0.90)
- Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection (NeurIPS 2026 · similarity 0.89)
- Structured Temporal Causality for Interpretable Multivariate Time Series Anomaly Detection (NeurIPS 2025 · similarity 0.89)
- When Model Meets New Normals: Test-Time Adaptation for Unsupervised Time-Series Anomaly Detection (AAAI 2024 · similarity 0.89)
- Detect, Explain, Interpret: An End-to-End Benchmark for Time Series Anomaly Detection, Explainability and Interpretability. (NeurIPS 2026 · 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.
