Data-driven design of complex network structures to promote synchronization

TL;DR

This paper introduces a data-driven method for designing complex network structures that enhance synchronization, especially when traditional optimization is infeasible due to uncertain node dynamics, demonstrating improved strategies through case studies.

Contribution

It proposes a novel data-driven approach for optimizing network interconnections to promote synchronization, incorporating node dynamics and combining neural networks with genetic algorithms.

Findings

Heterogeneous optimal graphs emerge when including node dynamics.

Data-driven strategies outperform traditional methods in synchronization tasks.

Neural network and genetic algorithm hybrid yields statistically superior designs.

Abstract

We consider the problem of optimizing the interconnection graphs of complex networks to promote synchronization. When traditional optimization methods are inapplicable, due to uncertain or unknown node dynamics, we propose a data-driven approach leveraging datasets of relevant examples. We analyze two case studies, with linear and nonlinear node dynamics. First, we show how including node dynamics in the objective function makes the optimal graphs heterogeneous. Then, we compare various design strategies, finding that the best either utilize data samples close to a specific Pareto front or a combination of a neural network and a genetic algorithm, with statistically better performance than the best examples in the datasets.