Wednesday, August 10, 2016

Exact Recovery of Chaotic Systems from Highly Corrupted Data

The following is a somewhat improvement over the recent Discovering governing equations from data: Sparse identification of nonlinear dynamical systems - implementation - ( see also the Video: Sparse Identification of Nonlinear Dynamics (SINDy)) as the recovery now takes into account (and removes) large outliers. This is from Rachel's group:




Exact Recovery of Chaotic Systems from Highly Corrupted Data by Giang Tran, Rachel Ward
Learning the governing equations in dynamical systems from time-varying measurements is of great interest across different scientific fields. This task becomes prohibitive when such data is moreover highly corrupted, for example, due to the recording mechanism failing over unknown intervals of time. When the underlying system exhibits chaotic behavior, such as sensitivity to initial conditions, it is crucial to recover the governing equations with high precision. In this work, we consider continuous time dynamical systems x˙=f(x) where each component of f:RdRd is a multivariate polynomial of maximal degree p; we aim to identify f exactly from possibly highly corrupted measurements x(t1),x(t2),,x(tm). As our main theoretical result, we show that if the system is sufficiently ergodic that this data satisfies a strong central limit theorem (as is known to hold for chaotic Lorenz systems), then the governing equations f can be exactly recovered as the solution to an 1 minimization problem -- even if a large percentage of the data is corrupted by outliers. Numerically, we apply the alternating minimization method to solve the corresponding constrained optimization problem. Through several examples of 3D chaotic systems and higher dimensional hyperchaotic systems, we illustrate the power, generality, and efficiency of the algorithm for recovering governing equations from noisy and highly corrupted measurement data.




 
Join the CompressiveSensing subreddit or the Google+ Community or the Facebook page and post there !
Liked this entry ? subscribe to Nuit Blanche's feed, there's more where that came from. You can also subscribe to Nuit Blanche by Email, explore the Big Picture in Compressive Sensing or the Matrix Factorization Jungle and join the conversations on compressive sensing, advanced matrix factorization and calibration issues on Linkedin.

No comments:

Printfriendly