Research map / Time series / Anomaly detection
Unsupervised Model Selection for Time Series 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 |
|---|---|---|---|
| Combine selected surrogate rankings with robust Borda, filtering rankings by empirical influence | Unsupervised time-series anomaly detector selection · 10 datasets | significant losses to Oracle: 6 vs 8–9 for the listed individual surrogate metrics and Borda | Helped |
| Use Robust Borda instead of Borda (ablation) | Surrogate-ranking aggregation · 10 datasets | significant wins over Random: 4 vs 3 | Helped |
| Use Robust Borda instead of Borda (ablation) | Surrogate-ranking aggregation · 10 datasets | significant losses to Oracle: 6 vs 9 | Helped |
| Use the single minimum-influence metric instead of aggregating all rankings (ablation) | Surrogate-ranking selection · 10 datasets | significant wins over Random: 7 vs 4 | Helped |
| Use synthetic anomaly injection metrics rather than prediction-error or centrality metrics | Unsupervised model selection · 10 datasets | adjusted best F1 of the selected model | Helped |
| Use model centrality rankings | Unsupervised model selection · 10 datasets | adjusted best F1 of the selected model | Hurt |
| Select by prediction error (MSE) | Unsupervised model selection · NASA, RESP, SMD; ECG, ABP, Acceleration | adjusted best F1 of the selected model | Mixed |
| Discard high empirical-influence rankings before Borda aggregation (ablation) | Robust Borda · 10 datasets | significant losses to Oracle: 6 vs 9 | Helped |
| Add the top-k restriction to Borda rankings (ablation) | Borda aggregation · 10 datasets | adjusted best F1 of the selected model | No clear effect |
| Increase noise in the input permutations | Borda rank aggregation on synthetic rankings · Synthetic data experiments | distance of median permutation from central permutation | Hurt |
| Inject synthetic noise anomalies for model selection | Air Temperature · Air Temperature | adjusted best F1 of the selected model | Helped |
| Inject speedup or cutoff anomalies for model selection | RESP · RESP | adjusted best F1 of the selected model | Helped |
Closest papers by meaning
- Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection (NeurIPS 2026 · similarity 0.89)
- Position: Quo Vadis, Unsupervised Time Series Anomaly Detection? (ICML 2024 · similarity 0.88)
- The Elephant in the Room: Towards A Reliable Time-Series Anomaly Detection Benchmark (NeurIPS 2024 · similarity 0.88)
- When Model Meets New Normals: Test-Time Adaptation for Unsupervised Time-Series Anomaly Detection (AAAI 2024 · similarity 0.87)
- SEAD: Unsupervised Ensemble of Streaming Anomaly Detectors (ICML 2025 · similarity 0.87)
- Supervision Recovery for Time Series Anomaly Detection via Counterfactual Pairing (NeurIPS 2026 · 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.