Figure 5
From: Accelerated crystal structure prediction of multi-elements random alloy using expandable features
Computational cost of generating a training dataset using DFT calculation. For the training set using raw features, we suppose that 80% of the training set is required from the original dataset. However, using expandable features, the cost for generating the training set of the binary alloy is only required because the training algorithm using expandable features can be directly applicable to the multi-element alloys.
