The population is then evolved using traditional GP evolutionary operators (see Fig.6). As it can be observed in Fig.7), the heuristic template takes a scoring function as a parameter. Therefore, the fitness value of a specific scoring function can then be obtained by solving all instances using the template parametrized by this same scoring function. The sum of all objective values obtained on the training set (see Fig.8) becomes the fitness value of the scoring function and measure its abilities to solve the training instances.