Showing 1–1 of 1 results for author: Razzanelli, M
-
Estimating the State of Epidemics Spreading with Graph Neural Networks
Authors:
Abhishek Tomy,
Matteo Razzanelli,
Francesco Di Lauro,
Daniela Rus,
Cosimo Della Santina
Abstract:
When an epidemic spreads into a population, it is often unpractical or impossible to have a continuous monitoring of all subjects involved. As an alternative, algorithmic solutions can be used to infer the state of the whole population from a limited amount of measures. We analyze the capability of deep neural networks to solve this challenging task. Our proposed architecture is based on Graph Con…
▽ More
When an epidemic spreads into a population, it is often unpractical or impossible to have a continuous monitoring of all subjects involved. As an alternative, algorithmic solutions can be used to infer the state of the whole population from a limited amount of measures. We analyze the capability of deep neural networks to solve this challenging task. Our proposed architecture is based on Graph Convolutional Neural Networks. As such it can reason on the effect of the underlying social network structure, which is recognized as the main component in the spreading of an epidemic. We test the proposed architecture with two scenarios modeled on the CoVid-19 pandemic: a generic homogeneous population, and a toy model of Boston metropolitan area.
△ Less
Submitted 10 May, 2021;
originally announced May 2021.
