Empirical and Statistical Analysis of Genetic Programming with Linear Genome

The e ciency of a GP system largely depends on the representation of GP individuals. In the substitution for conventional S-expression or pointer-based systems, Linear GP, in which a program structure is represented as a linear genome, was proposed. This paper's primary research topic is the implementation of a fast, low memory consuming and ADF capable Linear GP system. In order to show the e ectiveness, we statistically analyze the performance of the system with some benchmark experiments.