Figure 3
From: Surprise maximization reveals the community structure of complex networks
A simple meta-algorithm based on Surprise maximization improves over all known community detection algorithms.
(a) Performances (calculated as in Figure 2) for all the algorithms are compared in both the LFR and the RC benchmarks with the performance of a strategy that consists in picking up the algorithm that provides the highest S value (Smax). (b) For the Smax strategy, the average VI values for the 1200 networks analyzed are very close to zero, i.e. an almost optimal performance.
