Fig. 12

Results for model trained on 50,000 samples (\(X_{train} \in [-4,4]\)). The shaded area represents the epistemic uncertainty. a test set of 1000 samples uniformly sampled from \(X_{test} \in [-4;4]\), \(\rho _{rank}^{sim}\) = 0.27±0.03, b test set of 1000 samples uniformly sampled from \(X_{test} \in [-6;6]\) \(\rho _{rank}^{sim}\) = 0.65±0.02