Learning Stochastic Dynamics with Statistics-Informed Neural Network
arXiv:2202.12278 · doi:10.1016/j.jcp.2022.111819
Abstract
We introduce a machine-learning framework named statistics-informed neural network (SINN) for learning stochastic dynamics from data. This new architecture was theoretically inspired by a universal approximation theorem for stochastic systems, which we introduce in this paper, and the projection-operator formalism for stochastic modeling. We devise mechanisms for training the neural network model to reproduce the correct \emph{statistical} behavior of a target stochastic process. Numerical simulation results demonstrate that a well-trained SINN can reliably approximate both Markovian and non-Markovian stochastic dynamics. We demonstrate the applicability of SINN to coarse-graining problems and the modeling of transition dynamics. Furthermore, we show that the obtained reduced-order model can be trained on temporally coarse-grained data and hence is well suited for rare-event simulations.
References in corpus (11)
- PPINN: Parareal Physics-Informed Neural Network for time-dependent PDEs
- Quantifying total uncertainty in physics-informed neural networks for solving forward and inverse stochastic problems
- Wave Physics as an Analog Recurrent Neural Network
- Data-driven parameterization of the generalized Langevin equation
- SPINN: Sparse, Physics-based, and partially Interpretable Neural Networks for PDEs
- Generalized Fokker-Planck equation: Derivation and exact solutions
- Machine Learning for Prediction with Missing Dynamics
- Learning interaction kernels in heterogeneous systems of agents from multiple trajectories
- Existence and estimates of moments for Lévy-type processes
- Kernel-based Prediction of Non-Markovian Time Series
- On the Asymptotic Behavior of the Kernel Function in the Generalized Langevin Equation: A One-dimensional lattice model