Figure 6
From: Uncertainty-aware mixed-variable machine learning for materials design
(a,b) Relative RMSEs of fitting quadratic and Perm functions, using LVGP and Lolo models with varying training sample sizes. Boxplots show results from 10 different randomly selected training sample sets. (c,d) Regression plots showing the true value (horizontal axis) and ML model-predicted value (vertical axis) for quadratic function (training size 50) and Perm function (training size 100), with green diagonal lines representing accurate predictions.
