Semantic genetic operators based on a selection mechanism tailored for the root function

Genetic Programming (GP) is an evolutionary algorithm that has received a lot of attention because it has been successfully applied to solving hard real problems. Furthermore, it has been shown that the use of semantic operators can improve GP performance in supervised learning problems. In this work, for parents selection, target semantics and function's properties are used. We propose three new selection techniques: tournament selection based on desired semantics, tournament selection based on the orthogonality and tournament selection for one argument functions. To prove the performance of our proposal, we test our implementation with different classification problems.