Tailored forecasting from short time series via meta-learning
Science Advances · 17.07.2026 · 12 (29)
DOI: 10.1126/sciadv.ady7216 ↗Kurz gefasst
Fachartikel in der Zeitschrift Science Advances, erschienen am 17.07.2026. Verfasst von Declan A Norton und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Physics, University of Maryland, College Park, MD 20742, USA. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 86 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: Office of Naval Research, Defense Advanced Research Projects Agency, National Science Foundation Graduate Research Fellowship Program.
Abstract (Originalsprache)
Machine learning models can effectively forecast dynamical systems from time-series data, but they typically require large amounts of past data, making forecasting particularly challenging for systems with limited history. To overcome this, we introduce meta-learning for tailored forecasting using related time series (METAFORS), which generalizes knowledge across systems to enable forecasting in data-limited scenarios. By learning from a library of models trained on longer time series from potentially related systems, METAFORS builds and initializes a model tailored to short time-series data from the system of interest. Using a reservoir computing implementation and testing on simulated chaotic systems, we demonstrate that METAFORS can reliably predict both short-term dynamics and long-term statistics without requiring contextual labels. We see this even when test and related systems exhibit substantially different behaviors, highlighting METAFORS’ strengths in data-limited scenarios.
Autor:innen und Institutionen
- Department of Physics, University of Maryland, College Park, MD 20742, USA.
- Institute for Research in Electronics and Applied Physics, University of Maryland, College Park, MD 20742, USA.
- Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742, USA.
- Potomac Research LLC, Alexandria, VA 22314, USA.
- Institute for Physical Science and Technology, University of Maryland, College Park, MD 20742, USA.
- Department of Mathematics, University of Maryland, College Park, MD 20742, USA.
ORCID-Profile: Declan A Norton, Michelle Girvan
Förderung
- Office of Naval Research
- Defense Advanced Research Projects Agency
- National Science Foundation Graduate Research Fellowship Program
Zitieren
APA
Norton, D. A., Ott, E., Pomerance, A., Hunt, B. R., & Girvan, M. (2026). Tailored forecasting from short time series via meta-learning. Science Advances, 12(29). https://doi.org/10.1126/sciadv.ady7216
Vancouver
Norton DA, Ott E, Pomerance A, Hunt BR, Girvan M. Tailored forecasting from short time series via meta-learning. Science Advances. 2026;12(29). doi:10.1126/sciadv.ady7216
BibTeX
@article{Norton2026,
author = {Norton, Declan A and Ott, Edward and Pomerance, Andrew and Hunt, Brian R and Girvan, Michelle},
title = {Tailored forecasting from short time series via meta-learning},
journal = {Science Advances},
year = {2026},
volume = {12},
number = {29},
doi = {10.1126/sciadv.ady7216}
}
EUVIMED