Abstract:The discovery of governing differential equations from data is an open frontier in machine learning. The sparse identification of nonlinear dynamics (SINDy) \citep{brunton_discovering_2016} framework enables data-driven discovery of interpretable models in the form of sparse, deterministic governing laws. Recent works have sought to adapt this approach to the stochastic setting, though these adaptations are severely hampered by the curse of dimensionality. On the other hand, Bayesian-inspired deep learning methods have achieved widespread success in high-dimensional probabilistic modeling via computationally efficient approximate inference techniques, suggesting the use of these techniques for efficient stochastic equation discovery. Here, we introduce HyperSINDy, a framework for modeling stochastic dynamics via a deep generative model of sparse governing equations whose parametric form is discovered from data. HyperSINDy employs a variational encoder to approximate the distribution of observed states and derivatives. A hypernetwork \citep{ha_hypernetworks_2016} transforms samples from this distribution into the coefficients of a differential equation whose sparse form is learned simultaneously using a trainable binary mask \citep{louizos_learning_2018}. Once trained, HyperSINDy generates stochastic dynamics via a differential equation whose coefficients are driven by a Gaussian white noise. In experiments, HyperSINDy accurately recovers ground truth stochastic governing equations, with learned stochasticity scaling to match that of the data. Finally, HyperSINDy provides uncertainty quantification that scales to high-dimensional systems. Taken together, HyperSINDy offers a promising framework for model discovery and uncertainty quantification in real-world systems, integrating sparse equation discovery methods with advances in statistical machine learning and deep generative modeling.

Symbolic Regression on Sparse and Noisy Data with Gaussian Processes

Active data based Gaussian process models for nonlinear spatiotemporal systems

Rapid Bayesian identification of sparse nonlinear dynamics from scarce and noisy data

Gaussian Processes with Noisy Regression Inputs for Dynamical Systems

Deep Generative Modeling for Identification of Noisy, Non-Stationary Dynamical Systems

Sparse Identification of Nonlinear Dynamics with Side Information (SINDy-SI)

HyperSINDy: Deep Generative Modeling of Nonlinear Stochastic Governing Equations

A Robust SINDy Approach by Combining Neural Networks and an Integral Form

Generalizing the SINDy approach with nested neural networks

Learning sparse nonlinear dynamics via mixed-integer optimization

Gaussian process learning of nonlinear dynamics

CINDy: Conditional gradient-based Identification of Non-linear Dynamics -- Noise-robust recovery

Bayesian autoencoders for data-driven discovery of coordinates, governing equations and fundamental constants

Identifying Nonlinear Dynamics with High Confidence from Sparse Data

GN-SINDy: Greedy Sampling Neural Network in Sparse Identification of Nonlinear Partial Differential Equations

Cautious Model Predictive Control Using Gaussian Process Regression

Multi-objective SINDy for parameterized model discovery from single transient trajectory data

Scalable Sparse Regression for Model Discovery: The Fast Lane to Insight

Model selection for hybrid dynamical systems via sparse regression

A Variant of Gaussian Process Dynamical Systems.

Vehicle Dynamics Modeling for Autonomous Racing Using Gaussian Processes