Table 4 Prediction performance of logS and logBB tasks when the training data has the same number of samples
From: Improving VAE based molecular representations for compound property prediction
| Â | Dataset size | No descriptors | With the top 1 descriptor |
|---|---|---|---|
logS (R2) | 6668 | 0.745 | 0.796 |
logS (R2) | 2906 | 0.676 | 0.763 |
logBB (Accuracy/F1) | 2906 | 0.887/0.932 | 0.888/0.934 |