A study of the neutrality of Boolean function landscapes in genetic programming

The neutrality of genetic programming Boolean function landscapes is investigated in this paper. Compared with some well-known contributions on the same issue, (i) we first define new measures which help in characterizing neutral landscapes; (ii) we use a new sampling methodology, which captures features that are disregarded by uniform random sampling; (iii) we introduce new genetic operators to define the neighborhood of tree structures; and (iv) we compare the fitness landscape induced by different sets of functional operators. This study indicates the existence of a relationship between our neutrality measures and the performance of genetic programming for the problems studied.