Table 2 The performance of the Graph Transformer with different exploration rates in the RL framework
From: DrugEx v3: scaffold-constrained drug design with graph transformer-based reinforcement learning
ε | Accuracy | Desirability | Uniqueness | Novelty | Diversity |
|---|---|---|---|---|---|
0.0 | 99.7% | 74.6% | 60.7% | 60.6% | 0.879 |
0.1 | 99.7% | 66.8% | 75.0% | 74.6% | 0.842 |
0.2 | 99.8% | 61.6% | 80.2% | 79.4% | 0.879 |
0.3 | 99.7% | 56.8% | 89.8% | 88.8% | 0.874 |
0.4 | 99.7% | 54.8% | 88.8% | 87.5% | 0.859 |
0.5 | 99.7% | 46.8% | 88.5% | 86.4% | 0.875 |