Machine learning with data assimilation and uncertainty quantification for dynamical systems: a review

Sibo Cheng,Cesar Quilodran-Casas,Said Ouala,Alban Farchi,Che Liu,Pierre Tandeo,Ronan Fablet,Didier Lucor,Bertrand Iooss,Julien Brajard,Dunhui Xiao,Tijana Janjic,Weiping Ding,Yike Guo,Alberto Carrassi,Marc Bocquet,Rossella Arcucci

2023-03-19

Abstract:Data Assimilation (DA) and Uncertainty quantification (UQ) are extensively used in analysing and reducing error propagation in high-dimensional spatial-temporal dynamics. Typical applications span from computational fluid dynamics (CFD) to geoscience and climate systems. Recently, much effort has been given in combining DA, UQ and machine learning (ML) techniques. These research efforts seek to address some critical challenges in high-dimensional dynamical systems, including but not limited to dynamical system identification, reduced order surrogate modelling, error covariance specification and model error correction. A large number of developed techniques and methodologies exhibit a broad applicability across numerous domains, resulting in the necessity for a comprehensive guide. This paper provides the first overview of the state-of-the-art researches in this interdisciplinary field, covering a wide range of applications. This review aims at ML scientists who attempt to apply DA and UQ techniques to improve the accuracy and the interpretability of their models, but also at DA and UQ experts who intend to integrate cutting-edge ML approaches to their systems. Therefore, this article has a special focus on how ML methods can overcome the existing limits of DA and UQ, and vice versa. Some exciting perspectives of this rapidly developing research field are also discussed.

Machine Learning

What problem does this paper attempt to address?

The paper primarily explores the combined application of Machine Learning (ML), Data Assimilation (DA), and Uncertainty Quantification (UQ) in high-dimensional dynamical systems and provides a review of this topic. Specifically, the goals of the paper include: 1. **Addressing key challenges in high-dimensional dynamical systems**: These challenges include, but are not limited to, the identification of dynamical systems, the construction of reduced-order surrogate models, the specification of error covariance, and the correction of model errors. 2. **Providing a comprehensive guide**: Given that a large number of techniques and methods have been developed, this paper aims to offer a thorough overview of the cutting-edge research in this interdisciplinary field. 3. **Promoting communication and collaboration across different fields**: The paper is intended not only for machine learning scientists attempting to apply DA and UQ techniques to improve model accuracy and interpretability but also for DA and UQ experts looking to integrate advanced ML methods into their systems. 4. **Demonstrating how ML can overcome existing limitations of DA and UQ**: The paper particularly focuses on how ML methods can address the key challenges of DA and UQ, and vice versa. 5. **Discussing research prospects**: The paper also discusses the exciting research prospects and challenges faced in this rapidly developing field. The paper covers a wide range of application scenarios, such as Numerical Weather Prediction (NWP), environmental modeling, and Computational Fluid Dynamics (CFD). Additionally, the paper emphasizes that its goal is not to compare the performance of existing methods, as these methods were developed to address different problems. Overall, this review paper aims to provide readers with a comprehensive perspective on how to combine ML, DA, and UQ for application in high-dimensional dynamical systems.

Machine learning with data assimilation and uncertainty quantification for dynamical systems: a review

Uncertainty Quantification For Turbulent Flows with Machine Learning

Navigating Uncertainties in Machine Learning for Structural Dynamics: A Comprehensive Review of Probabilistic and Non-Probabilistic Approaches in Forward and Inverse Problems

Recent Advances on Machine Learning for Computational Fluid Dynamics: A Survey

A review of uncertainty quantification in deep learning: Techniques, applications and challenges

Evaluation of machine learning architectures on the quantification of epistemic and aleatoric uncertainties in complex dynamical systems

A Structured Review of Literature on Uncertainty in Machine Learning & Deep Learning

Multi-Fidelity Data Assimilation For Physics Inspired Machine Learning In Uncertainty Quantification Of Fluid Turbulence Simulations

Uncertainty Quantification in Machine Learning for Engineering Design and Health Prognostics: A Tutorial

Physics Based & Machine Learning Methods For Uncertainty Estimation In Turbulence Modeling

A Survey on Uncertainty Quantification Methods for Deep Learning

Combining data assimilation and machine learning to infer unresolved scale parametrisation

Quality measures for the evaluation of machine learning architectures on the quantification of epistemic and aleatoric uncertainties in complex dynamical systems

NeuralUQ: A Comprehensive Library for Uncertainty Quantification in Neural Differential Equations and Operators

Predictability of weakly turbulent systems from spatially sparse observations using data assimilation and machine learning

Uncertainty Quantification of Deep Learning for Spatiotemporal Data: Challenges and Opportunities

Surrogate modeling for the climate sciences dynamics with machine learning and data assimilation

Uncertainty Quantification in Computational Fluid Dynamics: Physics and Machine Learning Based Approaches

Review of Machine Learning for Hydrodynamics, Transport, and Reactions in Multiphase Flows and Reactors

Leveraging Viscous Hamilton–Jacobi PDEs for Uncertainty Quantification in Scientific Machine Learning

Applying Machine Learning Techniques: Uncertainty Quantification in Nonlinear Dynamics Characters Predictions via Gated Recurrent Unit-Based Reduced-Order Models