Statistical Inference Framework for Source Detection of Contagion Processes on Arbitrary Network Structures

Abstract

We introduce a statistical inference framework for maximum likelihood estimation of the contagion source from a partially observed contagion spreading process on an arbitrary network structure. The framework is based on simulations of a contagion spreading process from a set of potential sources which were infected in the observed realization. We present a number of different likelihood estimators for determining the conditional probabilities of potential initial sources producing the observed epidemic realization, which are computed in scalable and parallel way. This statistical inference framework is applicable to arbitrary networks with different dynamical spreading processes.