Figure 2.
Comparison results of OverfitDTI and baselines on the KIBA dataset in the scenario of warm start
(A–C) The predictive performance were evaluated with MSE (A) and CI (B) values. (C) Scatterplots of binding scores between the predicted values and the true values. OverfitDTI were trained with all data while the baselines were trained with 70% of all data.
