Fig. 8From: Assessing the information content of structural and protein–ligand interaction representations for the classification of kinase inhibitor binding modes via machine learning and active learningActive learning on the basis of chemical structure. Test set MCC (purple) and BA (blue) performance is shown for increasing numbers of training instances, with entropy-based (solid line) and random (dashed line) selection of compounds from the pool. Shaded areas of each curve indicate standard deviations of different prediction trialsBack to article page