Table 4.

Clustering performance on LFR-benchmark datasets.

Datasets	AP	GLay	MCL	SC	HC	mCODE	N2V-HC(MMS, DS, NPC)
LFR (100, 10, 30)	0.304	0.131	0.350	0.28	0.26	0.35	0.615 (6, 2, 8)
LFR (500, 10, 50)	0.090	0.127	0.120	0.128	0.14	0.138	0.496 (4, 3, 38)
LFR (1,000, 20, 100)	0.097	0.075	0.692	0.103	0.109	0.145	0.620 (6, 3, 40)
LFR (2,000, 30, 200)	0.092	0.033	0.651	0.080	0.082	0.135	0.682 (5, 2, 34)

AP, affinity propagation; MCL, Markov cluster; SC, spectral clustering; HC, hierarchical clustering; MMS, minModuleSize; DS, DeepSplit; NPC, number of predicted clusters. Parameter setting: MCL inflation factor was set in default (2.5) for all networks. Parameters in AP, GLay, SC, HC, and mCODE were in default except that cluster number was set to the ground truth if available. Bold Values indicate the best micro F1 scores.