Abstract:Experimental data is often affected by uncontrolled variables that make analysis and interpretation difficult. For spatiotemporal systems, this problem is further exacerbated by their intricate dynamics. Modern machine learning methods are particularly well-suited for analyzing and modeling complex datasets, but to be effective in science, the result needs to be interpretable. We demonstrate an unsupervised learning technique for extracting interpretable physical parameters from noisy spatiotemporal data and for building a transferable model of the system. In particular, we implement a physics-informed architecture based on variational autoencoders that is designed for analyzing systems governed by partial differential equations (PDEs). The architecture is trained end-to-end and extracts latent parameters that parameterize the dynamics of a learned predictive model for the system. To test our method, we train our model on simulated data from a variety of PDEs with varying dynamical parameters that act as uncontrolled variables. Numerical experiments show that our method can accurately identify relevant parameters and extract them from raw and even noisy spatiotemporal data (tested with roughly 10% added noise). These extracted parameters correlate well (linearly with $R^2 > 0.95$) with the ground truth physical parameters used to generate the datasets. We then apply this method to nonlinear fiber propagation data, generated by an ab-initio simulation, to demonstrate its capabilities on a more realistic dataset. Our method for discovering interpretable latent parameters in spatiotemporal systems will allow us to better analyze and understand real-world phenomena and datasets, which often have unknown and uncontrolled variables that alter the system dynamics and cause varying behaviors that are difficult to disentangle.

Discovering Nonlinear PDEs from Scarce Data with Physics-encoded Learning

Physics-Guided Discovery of Highly Nonlinear Parametric Partial Differential Equations

Physics-informed learning of governing equations from scarce data

Encoding physics to learn reaction–diffusion processes

Hard Encoding of Physics for Learning Spatiotemporal Dynamics

DL-PDE: Deep-learning based data-driven discovery of partial differential equations from discrete and noisy data

Physics-constrained robust learning of open-form partial differential equations from limited and noisy data

Predicting parametric spatiotemporal dynamics by multi-resolution PDE structure-preserved deep learning

Multi-resolution partial differential equations preserved learning framework for spatiotemporal dynamics

Extracting Interpretable Physical Parameters from Spatiotemporal Systems using Unsupervised Learning

Physics-informed machine learning

Learning a Neural Solver for Parametric PDE to Enhance Physics-Informed Methods

Physics-Based Deep Learning for Flow Problems

Incorporating Lasso Regression to Physics-Informed Neural Network for Inverse PDE Problem

Learning PDEs from data on closed surfaces with sparse optimization

Spatio-Temporal Super-Resolution of Dynamical Systems using Physics-Informed Deep-Learning

Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations

Physics-informed Spline Learning for Nonlinear Dynamics Discovery

Data-driven discovery of PDEs in complex datasets

Deep-learning based discovery of partial differential equations in integral form from sparse and noisy data