Towards a framework for observational causality from time series: when Shannon meets Turing

09/30/2019

∙

by   David Sigtermans, et al.

∙

0

∙

share

A tensor based formalism is proposed for inferring causal structures. This formalism enables us to determine the directionality of relations within a complex network. It furthermore allows us to differentiate between direct and indirect associations. Using this framework a Data Processing Inequality is proved to exist for Transfer Entropy.