Fig. 3
From: Module-based regularization improves Gaussian graphical models when observing noisy data
The optimal regularization strength as a function of the number of samples for Standard and Modular GLASSO. The Standard GLASSO regularizes less, resulting in the inclusion of many noisy correlations. The Modular GLASSO regularizes based on the modular structure, leading to a stronger regularization as the number of samples increases and the recovery of the planted modular structure. The large points represent averages over ten runs, with individual runs shown as small points. The AMI between recovered and planted partitions is displayed as a number next to each point