Table 2 Clustering performance evaluation using the 157 local features
From: Deep clustering of small molecules at large-scale via variational autoencoder embedding and K-means
Clustering method | #Clusters | Internal indices | ||
|---|---|---|---|---|
|  |  | Calinski–Harabasz | Silhouette | Davies–Bouldin |
K-means | 55 | 8509.651 | 0.124 | 1.704 |
BIRCH | 55 | 7243.245 | 0.082 | 1.831 |
VAE (16) + K-means | 105 | 7248.059 | 0.249 | 1.007 |
VAE (32) + K-means | 40 | 5166.671 | 0.197 | 1.194 |
VAE (64) + K-means | 35 | 9348.354 | 0.253 | 1.018 |
AE (16) + K-means | 30 | 4666.621 | 0.145 | 1.579 |
AE (32) + K-means | 50 | 5032.735 | 0.128 | 1.608 |
AE (64) + K-means | 50 | 5889.523 | 0.132 | 1.626 |