Fig. 4

Results on modularity optimization. In experiments, we suppose that the graph is partitioned into K communities with K ranging from 2 to 10 and report the maximum modularity value Q. We only perform experiments on two larger graphs: C.elegans and E-mail, since the sizes of karate and Jazz are too small. Experiment configuration: (GSO/EvoGSO): batch size = 256, initial \(\tau\) = 0.5, final \(\tau\) = 0.1, learning rate = 0.01, instance = 10, cycle \(T_1\) = 100, cycle \(T_2\) = 5000, substitution ratio 1/u = 1/8, mutation rate m = 0.001, elite ratio = 0.0625. (GA): population size = 64, crossover rate = 0.8, mutation rate = 0.001, and elite ratio = 0.125