Fig. 1
From: Splitting chemical structure data sets for federated privacy-preserving machine learning
Fold splitting in sparse multitask setting. For fold splitting in sparse multitask setting there are two approaches a Fold splitting is done on a per task level. For each task the observations are assigned to the test fold independently. In this setup, there are compounds which are for some assays in the test fold and for others in the training set. b Fold splitting on the whole compound space, where a compound with all its measurements is assigned to one fold