News ·
Wiener Filter on Linear Dynamic Graphs is local
We have shown that in a linear dynamic graph, where nodes represent stochastic processes and links represent LTI channels that model the influence of a node on another node, the Wiener filter and other filters are local. This insight can be used to reconstruct the topology of a network of stochastic processes connected via dynamical links. These results have interesting connections to machine learning and theory of causations.
Read more (opens in new window)
We have shown that in a linear dynamic graph, where nodes represent stochastic processes and links represent LTI channels that model the influence of a node on another node, the Wiener filter and other filters are local. This insight can be used to reconstruct the topology of a network of stochastic processes connected via dynamical links. These results have interesting connections to machine learning and theory of causations.