Simplifying functional network representation and interpretation through causality clustering

Functional networks, i.e. networks representing the interactions between the elements of a complex system and reconstructed from the observed elements’ dynamics, are becoming a fundamental tool to unravel the structures created by the movement of information in systems like the human brain. They also present drawbacks, one of the most important being the inherent difficulty in representing and interpreting the resulting structures for large number of nodes and links. I here propose a causality clustering approach, based on grouping nodes into clusters according to their similarity in the overall information dynamics, the latter one being measured by a causality metric. The whole system can then arbitrarily be simplified, with nodes being grouped in e.g. sources, brokers and sinks of information. The advantages and limitations of the proposed approach are discussed using a set of synthetic and real-world data sets, the latter ones representing two neuroscience and technological problems.This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement No 851255). The author acknowledges the Spanish State Research Agency, through the Severo Ochoa and María de Maeztu Program for Centers and Units of Excellence in R&D (MDM-2017-0711)