Table 6 Performance analysis of generated molecules considering the sampling approach of the training dataset
From: Designing optimized drug candidates with Generative Adversarial Network
Random | Tani-Inf | Tani-Sup | |
|---|---|---|---|
Validity (%) | 82.81 | 89.89 | 93.22 |
Uniqueness (%) | 96.89 | 94.73 | 97.63 |
Novelty (%) | 100.0 | 100.0 | 100.0 |
KL Divergence | 0.5266 | 0.2643 | 0.4915 |