Applying a sparse regression model in complex networks

With the growing availability of data from real systems in a plethora of scientific fields, a proportional expansion in the studies developed in the domain of complex networks hasallowed for the emergence of multiple analysis and prediction techniques in dynamical systems, as is done in the works involving information propagation. In parallel, and following anequally accelerated expansion rhythm, considerable advances have been achieved when applying machine learning methods, which is now an well-established field in the scientific community. In this work, we apply the Kuramoto model to generate synchronization patterns which will then generate a baseline for the SINDy model. The latter will be used to extract the governing equations of motion from a given network. The objective, which was to verify if SINDy can be effectively used to describe the time evolution of a given network, was achieved. Applying our methodology to known complex network models as well as to real networks, the analysis of the results with the metrics used demonstrated that this approach not only provides a reliable accuracy in uncovering and evolving the equations of motion obtained, but is also robust to the use of distinct types of networks.

Oct 17, 2022 — Oct 22, 2022

Applications of Nonlinear Systems to Socio-Economic Complexity