 |
Center for Nonlinear Studies |
Spreading processes play an increasingly important role in modeling for infectious diseases, regulatory networks, marketing and opinion setting. Events like Coronavirus or Cambridge Analytica affair further highlight the need for prediction, optimization, and control of diffusion dynamics. To tackle these tasks, it is essential to learn the effective spreading model and transmission probabilities across the network of interactions. Unfortunately, in most cases the transmission rates are unknown and need to be inferred from the spreading data. Additionally, it is rarely the case that we have full observation of the dynamics. As a result, typical maximum likelihood approach becomes intractable for large network instances. We introduce an efficient algorithm, based on the dynamic message-passing approach, which is able to reconstruct parameters of the effective spreading model given only limited information on the activation times of nodes in the network. The proposed method can be easily generalized to a large class of dynamic models on networks and beyond.
Host: Andrey Lokhov