Topology Identification under Spatially Correlated Noise

A bivariate zero-inflated Poisson control chart: Comments and corrections on earlier results

“…The topology estimation literature shows a variety of available methods to estimate the topology, such as Wiener filter based methods (Materassi & Innocenti, 2010;Materassi & Salapaka, 2012;Materassi, Salapaka, & Giarrè, 2011), Bayesian model selection techniques (Chiuso & Pillonetto, 2012;Shi, Bottegal, & Van den Hof, 2019;Wasserman, 2000), or methods that infer the topology from parametric estimates (Bolstad, Van Veen, & Nowak, 2011;Dankers, Van den Hof, Heuberger, & Bombois, 2012;Yuan, Stan, Warnick, & Gonçalves, 2011). While the main focus of topology detection literature has been on estimating network topology in the situation of a diagonal disturbance spectrum Φ v (ω), extensions towards nondiagonal spectra have been presented in Bombois and Hjalmarsson (2021), Dimovska and Materassi (2017) and Veedu and Salapaka (2020). In Veedu and Salapaka (2020) network topology and the non-zero pattern in the disturbance spectrum are estimated jointly.…”

“…While the main focus of topology detection literature has been on estimating network topology in the situation of a diagonal disturbance spectrum Φ v (ω), extensions towards nondiagonal spectra have been presented in Bombois and Hjalmarsson (2021), Dimovska and Materassi (2017) and Veedu and Salapaka (2020). In Veedu and Salapaka (2020) network topology and the non-zero pattern in the disturbance spectrum are estimated jointly. In this paper we assume that we do not know the disturbance topology a priori, but we assume that the network topology is known e.g., from its underlying physics, which is commonly the case for engineered systems.…”