Initialisation in Structured Grammatical Evolution

Robust initialisation has shown to greatly improve the performance of genetic programming on a wide variety of tasks. Many of these have been adapted to work with grammatical evolution, with varying success. We are the first to examine the effectiveness of some of the most popular grammatical evolution initialisation techniques using structured grammatical evolution. Namely, we investigate sensible initialisation and probabilistic tree creation 2, as well as the standard initialisation procedure used in structured grammatical evolution, grow. We examine their performance, as well as the diversity of solutions they create, on 7 well-known benchmarks. We observe that probabilistic tree creation 2 created the fittest initialisation populations on every benchmark considered. This did not result in overall better runs, however, and SGE runs with below average initialisation performance were seen to overcome their "bad start". The diversity of solutions, particularly fitness diversity, at the end of the run was lower for probabilistic tree creation 2 than for both sensible initialisation and grow.