Efficient and passive learning of networked dynamical systems driven by non-white exogenous inputs

We consider a networked linear dynamical system with agents/nodes. We study the problem of learning the underlying graph of interactions/dependencies from observations of the nodal trajectories over a time-interval . We present a regularized non-casual consistent estimator for this problem and analyze its sample complexity over two regimes: (a) where the interval consists of i.i.d. observation windows of length (restart and record), and (b) where is one continuous observation window (consecutive). Using the theory of -estimators, we show that the estimator recovers the underlying interactions, in either regime, in a time-interval that is logarithmic in the system size . To the best of our knowledge, this is the first work to analyze the sample complexity of learning linear dynamical systems \emph{driven by unobserved not-white wide-sense stationary (WSS) inputs}.
View on arXiv