Learning Stochastic Dynamical Systems via Bridge Sampling

We develop algorithms to automate discovery of stochastic dynamical system models from noisy, vector-valued time series. By discovery, we mean learning both a nonlinear drift vector field and a diagonal diffusion matrix for an Ito stochastic differential equation in \(\mathbb {R}^d\). We parameterize the vector field using tensor products of Hermite polynomials, enabling the model to capture highly nonlinear and/or coupled dynamics. We solve the resulting estimation problem using expectation maximization (EM). This involves two steps. We augment the data via diffusion bridge sampling, with the goal of producing time series observed at a higher frequency than the original data. With this augmented data, the resulting expected log likelihood maximization problem reduces to a least squares problem. We provide an open-source implementation of this algorithm. Through experiments on systems with dimensions one through eight, we show that this EM approach enables accurate estimation for multiple time series with possibly irregular observation times. We study how the EM method performs as a function of the amount of data augmentation, as well as the volume and noisiness of the data.

[1]  Paris Perdikaris,et al.  Machine learning of linear differential equations using Gaussian processes , 2017, J. Comput. Phys..

[2]  Steven L. Brunton,et al.  Inferring Biological Networks by Sparse Identification of Nonlinear Dynamics , 2016, IEEE Transactions on Molecular, Biological and Multi-Scale Communications.

[3]  T. Faniran.  Numerical Solution of Stochastic Differential Equations , 2015 .

[4]  Harry van Zanten,et al.  Reversible jump MCMC for nonparametric drift estimation for diffusion processes , 2012, Comput. Stat. Data Anal..

[5]  Harish S. Bhat,et al.  A Block Coordinate Descent Proximal Method for Simultaneous Filtering and Parameter Estimation , 2018, ICML.

[6]  J N Kutz,et al.  Model selection for dynamical systems via sparse regression and information criteria , 2017, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[7]  G. Roberts,et al.  Data Augmentation for Diffusions , 2013 .

[8]  Hayden Schaeffer,et al.  Extracting Sparse High-Dimensional Dynamics from Limited Data , 2017, SIAM J. Appl. Math..

[9]  M. Opper,et al.  Variational estimation of the drift for stochastic differential equations from the empirical density , 2016, 1603.01159.

[10]  Gareth O. Roberts,et al.  Importance sampling techniques for estimation of diffusion models , 2009 .

[11]  Dan Cornford,et al.  Variational Inference for Diffusion Processes , 2007, NIPS.

[12]  S. Brunton,et al.  Discovering governing equations from data by sparse identification of nonlinear dynamical systems , 2015, Proceedings of the National Academy of Sciences.

[13]  G. Roberts,et al.  On inference for partially observed nonlinear diffusion models using the Metropolis–Hastings algorithm , 2001 .

[14]  João Nicolau.  NONPARAMETRIC ESTIMATION OF SECOND-ORDER STOCHASTIC DIFFERENTIAL EQUATIONS , 2007, Econometric Theory.

[15]  S. Osher,et al.  Sparse dynamics for partial differential equations , 2012, Proceedings of the National Academy of Sciences.

[16]  Giang Tran,et al.  Exact Recovery of Chaotic Systems from Highly Corrupted Data , 2016, Multiscale Model. Simul..

[17]  B. Øksendal.  Stochastic differential equations : an introduction with applications , 1987 .

[18]  Nicolas Verzelen,et al.  Inferring stochastic dynamics from functional data , 2012 .

[19]  Ali Shojaie,et al.  Network Reconstruction From High-Dimensional Ordinary Differential Equations , 2016, Journal of the American Statistical Association.

[20]  Dan Cornford,et al.  Variational mean-field algorithm for efficient inference in large systems of stochastic differential equations. , 2015, Physical review. E, Statistical, nonlinear, and soft matter physics.

[21]  Fredrik Lindsten,et al.  Probabilistic learning of nonlinear dynamical systems using sequential Monte Carlo , 2017, ArXiv.

[22]  Steven L Brunton,et al.  Sparse identification of nonlinear dynamics for rapid model recovery. , 2018, Chaos.

[23]  Andreas Ruttor,et al.  Approximate Gaussian process inference for the drift function in stochastic differential equations , 2013, NIPS.

[24]  Steven L. Brunton,et al.  Data-driven discovery of partial differential equations , 2016, Science Advances.

[25]  Shagun Rawat,et al.  Learning governing equations for stochastic dynamical systems , 2018 .

[26]  Harish S. Bhat,et al.  Nonparametric Adjoint-Based Inference for Stochastic Differential Equations , 2016, 2016 IEEE International Conference on Data Science and Advanced Analytics (DSAA).

[27]  Harry van Zanten,et al.  Guided proposals for simulating multi-dimensional diffusion bridges , 2013, 1311.3606.

[28]  Andreas Ruttor,et al.  Approximate Bayes learning of stochastic differential equations. , 2017, Physical review. E.

[29]  I. G. MacKenzie,et al.  Stochastic Processes with Applications , 1992 .

[30]  Andrew M. Stuart,et al.  Inverse problems: A Bayesian perspective , 2010, Acta Numerica.

[31]  George E. Karniadakis,et al.  Hidden physics models: Machine learning of nonlinear partial differential equations , 2017, J. Comput. Phys..

[32]  H. Schaeffer,et al.  Learning partial differential equations via data discovery and sparse optimization , 2017, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences.

[33]  Moritz Schauer,et al.  Adaptive nonparametric drift estimation for diffusion processes using Faber–Schauder expansions , 2016 .

[34]  Zoubin Ghahramani,et al.  Learning Nonlinear Dynamical Systems Using an EM Algorithm , 1998, NIPS.

[35]  Hans-Georg Muller,et al.  Empirical dynamics for longitudinal data , 2010, 1211.2630.