Figure 3.

HEAD-DT in action. Figure 3 presents an illustration of HEAD-DT — a set of linear-genome individuals are decoded into algorithms and executed over the meta-training set. The performance of these individuals is measured and the evolutionary cycle starts with typical operations such as selection, crossover, and mutation. At the end of the cycle, the best (fitness-wise) individual is selected for inducing decision trees from the meta-test set.