Online learning of neural state-space models
2026-07-20 • Machine Learning
Machine Learning
AI summaryⓘ
The authors worked on improving how neural networks can model systems that change over time. They focused on making these models learn and update themselves continuously while new data comes in, instead of just learning once from a fixed batch of data. They created a new method that learns in small batches and updates the model step-by-step, which is faster and still accurate. They tested their approach with simulations and showed it works well for real-time applications.
deep learningnonlinear system identificationneural state-space modelsencoder-based estimationrecursive algorithmonline learningbatch-wise learningmultiple shootingmodel convergencesimulation studies
Authors
Bendegúz Györök, Tamás Péni, Maarten Schoukens, Roland Tóth
Abstract
Recent advances in deep-learning-based nonlinear system identification have led to encoder-based estimation of neural state-space (ANN-SS) models that achieve state-of-the-art performance in offline settings by estimating initial model states from past input-output data. These methods are typically used in multiple-shooting-based offline identification, and online learning of these models remains largely unexplored. This paper presents a batch-wise learning pipeline and a direct recursive identification algorithm for subspace encoder-based ANN-SS models. We provide convergence analysis of the recursive formulation and validate its performance through extensive simulation studies. The results demonstrate that the proposed approach enables computationally efficient online adaptation with high model accuracy.