Summary
We consider the problem of optimizing the interconnection graphs of complex networks to enhance synchronization. In instances where traditional optimization methods prove impractical due to uncertain or unknown node dynamics, we advocate for a data-driven approach that exploits available datasets containing pertinent examples. We consider two representative case studies, one featuring linear and the other nonlinear node dynamics. We explored different design strategies and discovered that the most effective ones either rely on using knowledge from examples close to a specific Pareto front or employ a combination of a neural network and a genetic algorithm. Notably, these strategies exhibit statistically superior performance compared to the best examples within the datasets.