What makes a small-world network? Leveraging machine learning for the robust prediction and classification of networks

Real-world network data derived from physical systems such as ecological food webs, biochemical pathways, genetic interactions, animal social behavior, and biological processes, captures complex relationships and addresses fundamental questions about species adaptability, ecosystem dynamics, pathogen dynamics, social dynamics, and genetic regulatory networks [3, 10, 18, 19, 29, 34]. The multi-dimensional nature and dynamic interactions among variables over time in these systems pose a challenge to their classification. Traditional classification methods (such as decision trees, support vector machines, k-nearest neighbor, and logistic regression) struggle to capture these complexities effectively [2, 27, 48, 52].