Detection of Topology Changes in Dynamical System: An Information Theoretic Approach
Résumé
In this paper, we show that the theory of information offers some tools to detect changes in the interaction topology of a dynamical system defined on a graph. As an illustrative example, the system we consider is a probabilistic voter model defined on a scale-free network. We show that, using time-delayed mutual-information, the interaction topology of an unknown graph can be reconstructed to some level. We apply this approach on a sliding time window to detect possible changes in the interaction topology over time.
Origine | Fichiers produits par l'(les) auteur(s) |
---|