Figure 1. Adjusted Rand indexes between the reference community structure used to generate the random multiplex networks and the communities detected by the different approaches.

Multiplex networks contain from 1 to 9 graph layers. The indexes are averaged over 1,000 random multiplex networks of 1,000 vertices and 20 balanced communities. Sparse (resp. dense) multiplex networks are simulated with 0.1/0.01 (resp. 0.5/0.2) internal/external edge probability matrix. Mixed multiplex networks are simulated by uniformly sampling among these two matrices. Each vertex is withdrawn from each graph with a probability 0.5 to generate missing data.