Feature engineering with regularity structures
From MaRDI portal
Abstract: We investigate the use of models from the theory of regularity structure as features in machine learning tasks. A model is a multi-linear function of a space-time signal designed to well-approximate solutions to partial differential equations (PDEs), even in low regularity regimes. Models can be seen as natural multi-dimensional generalisations of signatures of paths; our work therefore aims to extend the recent use of signatures in data science beyond the context of time-ordered data. We provide a flexible definition of a model feature vector associated to a space-time signal, along with two algorithms which illustrate ways in which these features can be combined with linear regression. We apply these algorithms in several numerical experiments designed to learn solutions to PDEs with a given forcing and boundary data. Our experiments include semi-linear parabolic and wave equations with forcing, and Burgers' equation with no forcing. We find an advantage in favour of our algorithms when compared to several alternative methods. Additionally, in the experiment with Burgers' equation, we noticed stability in the prediction power when noise is added to the observations.
Recommendations
- Learning partial differential equations via data discovery and sparse optimization
- A theory of regularity structures
- IDENT: identifying differential equations with numerical time evolution
- Deep hidden physics models: deep learning of nonlinear partial differential equations
- Sparse Deep Neural Network for Nonlinear Partial Differential Equations
Cites work
- A kernel two-sample test
- A theory of regularity structures
- Algebraic renormalisation of regularity structures
- An Augmented Lagrangian Deep Learning Method for Variational Problems with Essential Boundary Conditions
- An introduction to computational stochastic PDEs
- Characteristic functions of measures on geometric rough paths
- DGM: a deep learning algorithm for solving partial differential equations
- Deep Nitsche Method: Deep Ritz Method with Essential Boundary Conditions
- Differential equations driven by rough signals
- Dynamic mode decomposition of numerical and experimental data
- Extended Physics-Informed Neural Networks (XPINNs): A Generalized Space-Time Domain Decomposition Based Deep Learning Framework for Nonlinear Partial Differential Equations
- Integration of Paths--A Faithful Representation of Paths by Noncommutative Formal Power Series
- Kernels for sequentially ordered data
- Learning data-driven discretizations for partial differential equations
- Machine learning for fluid mechanics
- Numerical Method for Model-free Pricing of Exotic Derivatives in Discrete Time Using Rough Path Signatures
- Optimal execution with rough path signatures
- Parameter estimation for rough differential equations
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Regression modeling strategies. With applications to linear models, logistic regression, and survival analysis
- Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
- Solving high-dimensional partial differential equations using deep learning
- Spectral analysis of nonlinear flows
- Stochastic Equations in Infinite Dimensions
- The signature of a rough path: uniqueness
- Uniqueness for the signature of a path of bounded variation and the reduced path group
This page was built for publication: Feature engineering with regularity structures
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6184277)
