Online model learning with data-assimilated reservoir computers

2025-04-23

AI Summary

Addressing the challenge of online prediction for nonlinear spatiotemporal fields—exemplified by cylinder wake flows governed by the Navier–Stokes equations—this paper proposes a unified estimation framework integrating Proper Orthogonal Decomposition (POD) for dimensionality reduction, generalized autoregressive reservoir computing, and ensemble sequential data assimilation. For the first time, it enables simultaneous online estimation of the physical state, reservoir hidden states, and model parameters—eliminating reliance on pre-trained models and enabling robust adaptive learning from partial initialization. The method unifies Bayesian state estimation with ensemble Kalman filter–type assimilation strategies, supporting both full-field projection and sparse-sensor observation scenarios. In numerical experiments on cylinder wake flow, the triple estimation reduces reconstruction error by over 30% compared to dual estimation, while significantly improving ensemble convergence and real-time modeling capability.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Learning with Manifolds
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

We propose an online learning framework for forecasting nonlinear spatio-temporal signals (fields). The method integrates (i) dimensionality reduction, here, a simple proper orthogonal decomposition (POD) projection; (ii) a generalized autoregressive model to forecast reduced dynamics, here, a reservoir computer; (iii) online adaptation to update the reservoir computer (the model), here, ensemble sequential data assimilation.We demonstrate the framework on a wake past a cylinder governed by the Navier-Stokes equations, exploring the assimilation of full flow fields (projected onto POD modes) and sparse sensors. Three scenarios are examined: a na""ive physical state estimation; a two-fold estimation of physical and reservoir states; and a three-fold estimation that also adjusts the model parameters. The two-fold strategy significantly improves ensemble convergence and reduces reconstruction error compared to the na""ive approach. The three-fold approach enables robust online training of partially-trained reservoir computers, overcoming limitations of a priori training. By unifying data-driven reduced order modelling with Bayesian data assimilation, this work opens new opportunities for scalable online model learning for nonlinear time series forecasting.

Problem

• Online learning for nonlinear spatio-temporal signal forecasting
• Integration of dimensionality reduction and reservoir computing
• Online adaptation using ensemble sequential data assimilation

Innovation

• Online learning with reservoir computers
• Dimensionality reduction via POD projection
• Ensemble sequential data assimilation