Fig. 1From: Understanding the limitations of network online learningWe present the conditions under which network online learning is feasible. In particular, there is a potential for learning a policy to reduce incompleteness of a network only if the network has heterogenous degree distribution and high modularity. The synthetic graphs in the figure are ER (Erdös-Rényi), BA (Barabási-Albert), BTER (Block Two-level Erdös-Rényi), and LFR (Lancichinetti-Fortunato- Radicchi)Back to article page