EUVIMEDEUVIMED entdecken
Crossref

AL-KalmanNet: Neural Network-Aided Kalman Filtering Based on Active Learning

Andi Lin, Jianming Ruan, Wen-An Zhang

IEEE Signal Processing Letters · 2026 · 33 · 2370-2374

DOI: 10.1109/lsp.2026.3698385 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Signal Processing Letters, 2026. Verfasst von Andi Lin und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 32 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Neural network-aided Kalman filters (KFs) can learn from data to reliably estimate states in dynamic systems with complex and partially known dynamics. A key characteristic of current neural network-aided KF designs is their reliance on large amounts of labeled data. However, acquiring suitable datasets with a sufficient number of labeled data points consumes a significant workforce, making it impractical for real-world applications. Thus, we propose Active Learning-KalmanNet (AL-KalmanNet), a neural network-aided KF that integrates recent advances in deep learning-based filtering into the active learning framework in a practical manner. In AL-KalmanNet, an uncertainty-based acquisition function is developed to select the most informative unlabeled data points for querying, enabling neural network-aided KFs to learn from small amounts of labeled data and achieve performance comparable to that of neural network-aided KFs trained with sufficient data. Numerical simulations demonstrate the effectiveness of the proposed method.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Lin, A., Ruan, J., & Zhang, W. (2026). AL-KalmanNet: Neural Network-Aided Kalman Filtering Based on Active Learning. IEEE Signal Processing Letters, 33, 2370-2374. https://doi.org/10.1109/lsp.2026.3698385

Vancouver

Lin A, Ruan J, Zhang W. AL-KalmanNet: Neural Network-Aided Kalman Filtering Based on Active Learning. IEEE Signal Processing Letters. 2026;33:2370-2374. doi:10.1109/lsp.2026.3698385

BibTeX
@article{Lin2026,
  author = {Lin, Andi and Ruan, Jianming and Zhang, Wen-An},
  title = {AL-KalmanNet: Neural Network-Aided Kalman Filtering Based on Active Learning},
  journal = {IEEE Signal Processing Letters},
  year = {2026},
  volume = {33},
  pages = {2370--2374},
  doi = {10.1109/lsp.2026.3698385}
}
Kostenlos starten →