Combining Geometric Semantic GP with Gradient-Descent Optimization

Geometric semantic genetic programming (GSGP) is a well-known variant of (GP) where recombination and mutation operators have clear effect. Both kind randomly selected parameters that are not optimized by the search process. In this paper we combine GSGP with gradient-based optimizer, Adam, in order to leverage ability GP operate structural changes individuals methods optimize given structure. Two methods, named HYB-GSGP HeH-GSGP, defined compared on large set regression problems, showing use Adam can improve performance test set. The idea merging evolutionary computation optimization promising way combining two very different – complementary strengths.