Table 5 Evaluation of molecular diversity throughout the training process applying three distance measure
From: Diversity oriented Deep Reinforcement Learning for targeted molecule generation
Training step | 0 | 150 | 300 | 450 |
Tanimoto distance | 0.901 | 0.877 | 0.857 | 0.824 |
Levenshtein distance | 31.2 | 26.8 | 24.3 | 22.1 |
MCS distance | 0.799 | 0.767 | 0.724 | 0.713 |