New geometric semantic operators in genetic programming: perpendicular crossover and random segment mutation

Various geometric search operators have been developed to explore the behaviours of individuals in genetic programming (GP) for the sake of making the evolutionary process more effective. This work proposes two geometric search operators to fulfil the semantic requirements under the theoretical framework of geometric semantic GP for symbolic regression. The two operators approximate the target semantics gradually but effectively. The results show that the new geometric operators can not only lead to a notable benefit to the learning performance, but also improve the generalisation ability of GP. In addition, they also bring a significant improvement to Random Desired Operator, which is a state-of-the-art geometric semantic operator.