Generalized Jordan Center: A Source Localization Heuristic for Noisy and Incomplete Observations

Abstract

We consider estimating the source of a spreading process on a directed acyclic graph using noisy and incomplete observations; we believe this is the first work on source estimation under noisy information. Our main contribution is a novel heuristic, the generalized Jordan center (GJC), which is the maximum likelihood estimator of the diffusion source under mild conditions.