Published on Tue Feb 26 2019

Learning Dynamical Systems from Partial Observations

Ibrahim Ayed, Emmanuel de Bézenac, Arthur Pajot, Julien Brajard, Patrick Gallinari

We consider the problem of forecasting complex, nonlinear space-time processes. We propose a natural data-driven framework, where the system's dynamics are modelled by an unknown time-varying differential equation, and the Evolution term is estimated from the data.

0
0
0
Abstract

We consider the problem of forecasting complex, nonlinear space-time processes when observations provide only partial information of on the system's state. We propose a natural data-driven framework, where the system's dynamics are modelled by an unknown time-varying differential equation, and the evolution term is estimated from the data, using a neural network. Any future state can then be computed by placing the associated differential equation in an ODE solver. We first evaluate our approach on shallow water and Euler simulations. We find that our method not only demonstrates high quality long-term forecasts, but also learns to produce hidden states closely resembling the true states of the system, without direct supervision on the latter. Additional experiments conducted on challenging, state of the art ocean simulations further validate our findings, while exhibiting notable improvements over classical baselines.

Thu Jun 20 2019
Machine Learning
Data-driven prediction of a multi-scale Lorenz 96 chaotic system using deep learning methods: Reservoir computing, ANN, and RNN-LSTM
Three deep learning methods are examined. RC-ESN, deep feed-forward Artificial neural network (ANN), and recurrent neural network with long-term memory (RNN-LSTM) This Lorenz 96 system has three tiers of interacting variables.
0
0
0
Thu Aug 12 2021
Machine Learning
ST-PCNN: Spatio-Temporal Physics-Coupled Neural Networks for Dynamics Forecasting
A spatio-temporal physics-coupled neural network (ST-PCNN) model is proposed to achieve three goals: (1) learning the underlying physics, (2) transition of local information, and (3) forecasting future values for the dynamical system. The proposed
1
1
0
Sun May 02 2021
Machine Learning
Model discovery in the sparse sampling regime
0
0
0
Tue Nov 28 2017
Artificial Intelligence
Physics Informed Deep Learning (Part II): Data-driven Discovery of Nonlinear Partial Differential Equations
We introduce physics informed neural networks -- neural networks that are trained to solve supervised learning tasks while respecting any given law of physics. The effectiveness of our approach is demonstrated using a range of benchmark problems in mathematical physics.
0
0
0
Fri Oct 11 2019
Machine Learning
Predicting dynamical system evolution with residual neural networks
A data-driven approach to learning the evolution of dynamical systems is considered. We show how by training neural networks with ResNet-like architecture, models can be developed to predict the ODE system solution further in time.
0
0
0
Wed Feb 20 2019
Machine Learning
Shallow Neural Networks for Fluid Flow Reconstruction with Limited Sensors
Shallow neural network-based learning methodology for fluid flow reconstruction. No prior knowledge is assumed to be available, and the estimation method is purely data-driven.
0
0
0