EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Tackling data heterogeneity in federated time series forecasting

Wei Yuan, Chaoqun Yang, Xiangyu Zhao, Quoc Viet Hung Nguyen, Yang Cao, Tieke He, Hongzhi Yin

Science China Information Sciences · 2026-05 · 69 (5)

DOI: 10.1007/s11432-025-4553-x ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science China Information Sciences, online erschienen am 08.04.2026. Verfasst von Wei Yuan und 6 weiteren Autor:innen. Die Arbeit wurde bisher 6-mal zitiert (Crossref) und nennt 60 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Time series forecasting plays a critical role in various real-world applications, including energy consumption prediction, disease transmission monitoring, and weather forecasting. Although substantial progress has been made in time series forecasting, most existing methods rely on a centralized training paradigm, where large amounts of data are collected from distributed devices (e.g., sensors, wearables) to a central cloud server. However, this paradigm has overloaded communication networks and raised privacy concerns. Federated learning, a popular privacy-preserving technique, enables collaborative model training across distributed data sources. However, directly applying federated learning to time series forecasting often yields suboptimal results, as time series data generated by different devices is inherently heterogeneous. In this paper, we propose a novel framework, federated time series forecasting with synthetic data (Fed-TREND), to address data heterogeneity by generating informative synthetic data as auxiliary knowledge carriers. Specifically, Fed-TREND generates two types of synthetic data. The first type of synthetic data captures the representative distribution information from clients’ uploaded model updates and enhances clients’ local training consensus. The second kind of synthetic data extracts long-term influence insights from global model update trajectories and is used to refine the global model after aggregation. Fed-TREND is compatible with most time series forecasting models and can be seamlessly integrated into existing federated learning frameworks to improve prediction performance. Extensive experiments on eight datasets, using several federated learning baselines and four popular time series forecasting models, demonstrate the effectiveness and generalizability of Fed-TREND.

Zitieren

APA

Yuan, W., Yang, C., Zhao, X., Nguyen, Q. V. H., Cao, Y., He, T., & Yin, H. (2026). Tackling data heterogeneity in federated time series forecasting. Science China Information Sciences, 69(5). https://doi.org/10.1007/s11432-025-4553-x

Vancouver

Yuan W, Yang C, Zhao X, Nguyen QVH, Cao Y, He T, et al. Tackling data heterogeneity in federated time series forecasting. Science China Information Sciences. 2026;69(5). doi:10.1007/s11432-025-4553-x

BibTeX
@article{Yuan2026,
  author = {Yuan, Wei and Yang, Chaoqun and Zhao, Xiangyu and Nguyen, Quoc Viet Hung and Cao, Yang and He, Tieke and Yin, Hongzhi},
  title = {Tackling data heterogeneity in federated time series forecasting},
  journal = {Science China Information Sciences},
  year = {2026},
  volume = {69},
  number = {5},
  doi = {10.1007/s11432-025-4553-x}
}
Kostenlos starten →