Table 3.

Average execution time of clustering algorithms

	KMeans	Leidena	Birch	GMM	MeanShifta	Spectral Clustering	Hierarchical Clustering
Avg execution time	0.14	7.94	0.44	0.46	14.01	0.78	0.48
Speed factor relative to KMeans	1	56.71	3.17	3.29	100.09	5.60	3.47

We measured the mean duration across 3 consecutive runs on clustering the embedding produced by contrastive-sc for all real-world datasets. The methods annotated with ^aare those where the correct number of clusters was not provided as input. KMeans provides the highest computational speed from the set of 7 explored clustering algorithms

The fastest methods are highlighted in bold