Inverse based input estimation in stochastic networks


The goal of this project is to develop and analyze statistical inference methods for networks driven by stochastic processes. A complication that one typically faces is that the stochastic process feeding the network is not observed directly, and therefore indirect estimation techniques are needed. In a quintessential example, from snapshots of the network population one wishes to estimate the network's input characteristics. While for specific models results have been established, this area still offers a broad array of challenging open questions. The project lies at the intersection of applied probability and mathematical statistics. Potential application areas include service systems and communication networks.

Supervisors Michel Mandjes (UvA), Liron Ravner (UHaifa)
PhD Student Boris Lebedenko (UvA)
Location University of Amsterdam (UvA)