Research map / Time series / Causal and generative time series
Merlion: End-to-End Machine Learning for Time Series
Aadyot Bhatnagar, Paul Kassianik, Chenghao Liu, Tian Lan, Wenzhuo Yang, Rowan Cassius, Doyen Sahoo, Devansh Arpit, Sri Subramanian, Gerald Woo, Amrita Saha, Arun Kumar Jagota and 11 more
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 |
|---|---|---|---|
| add AutoML hyperparameter selection | ARIMA (AutoSARIMA) · m4-hour | mean sMAPE: 33.54 -> 13.61 | Improved |
| add AutoML hyperparameter selection | ARIMA (AutoSARIMA) · m4-year | mean sMAPE: 16.37 -> 17.16 | Hurt |
| add AutoML hyperparameter selection | Prophet (AutoProphet) · m4-hour | mean sMAPE: 18.08 -> 16.49 | Improved |
| add AutoML hyperparameter selection | Prophet (AutoProphet) · internal3 | mean sMAPE: 72.55 -> 69.32 | Improved |
| add AutoML hyperparameter selection | Prophet (AutoProphet) · internal1 | mean sMAPE: 56.07 -> 56.43 | Hurt |
| add AutoML hyperparameter selection | ETS (AutoETS) · m4-hour | mean sMAPE: 42.95 -> 19.23 | Improved |
| add AutoML hyperparameter selection | ETS (AutoETS) · internal1 | mean sMAPE: 17.02 -> 21.65 | Hurt |
| add AutoML hyperparameter selection | ETS (AutoETS) · m4-day | mean sMAPE: 3.04 -> 3.07 | No clear effect |
| ensemble calibrated scores of AutoETS, RRCF and ZMS instead of the single AutoETS | anomaly detector · NAB | F1: 0.350 -> 0.548 | Improved |
| ensemble calibrated scores of AutoETS, RRCF and ZMS instead of the single RRCF | anomaly detector · UCR | F1: 0.568 -> 0.476 | Hurt |
| ensemble calibrated scores of AutoETS, RRCF and ZMS instead of the single ZMS | anomaly detector · internal | F1: 0.486 -> 0.500 | No clear effect |
| ensemble calibrated scores instead of ARIMA | anomaly detector · average gap to best F1 | delta F1 vs best: 0.148 ± 0.099 -> 0.034 ± 0.044 | Improved |
Closest papers by meaning
- Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection (NeurIPS 2026 · similarity 0.86)
- Can LLMs Understand Time Series Anomalies? (ICLR 2025 · similarity 0.85)
- The Elephant in the Room: Towards A Reliable Time-Series Anomaly Detection Benchmark (NeurIPS 2024 · similarity 0.85)
- SEMPO: Lightweight Foundation Models for Time Series Forecasting (NeurIPS 2025 · similarity 0.85)
- ChronoEpilogi: Scalable Time Series Selection with Multiple Solutions (NeurIPS 2024 · similarity 0.85)
- When Model Meets New Normals: Test-Time Adaptation for Unsupervised Time-Series Anomaly Detection (AAAI 2024 · similarity 0.85)
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.
