Neural Gas Clustering Adapted for Given Size of Clusters

Table 2

Best MSE, mean MSE, and standard deviation of MSE of 100 runs for distances of data points from centers of clusters.

Dataset	Size	Clusters	Algorithm	Best	Mean	St. dev.
S1 subset	500	15	Balanced k-means	3.48	3.73	0.21
			Adapted neural gas	3.48	3.56	0.04
att532	532	4	Balanced k-means	1.597	1.597	0
			Adapted neural gas	1.597	1.597	1.0 × 10^−4
att532	532	7	Balanced k-means	0.723	0.723	8.6 × 10^−5
			Adapted neural gas	0.723	0.723	1.0 × 10^−4
att532	532	14	Balanced k-means	0.343	0.351	5.9 × 10^−3
			Adapted neural gas	0.343	0.343	1.2 × 10^−4
att532	532	19	Balanced k-means	0.252	0.266	0.012
			Adapted neural gas	0.252	0.252	1.6 × 10^−4