Keeping the Diversity with Small Populations Using Logic-Based Genetic Programming

We present a new method of Logic-Based Genetic Programming (LBGP). Using the intrinsic mechanism of backtracking in Prolog, we utilize large individual programs with redundant clauses, and apply them to small populations. Our method is validated by the experiments about Symbolic Regression, and XOR problems. Genetic programming is a method to automate programming tasks in which programs are evolved through Genetic Algorithm (GA) techniques. One of the difficulties of GP [2] is that I) they must have a large number of individuals in order to keep the population diversity, and that II) they cannot reuse sub-structures of program codes destroyed by improper uses of genetic operators. To cope with the issues, we will focus on the Prolog-based logic programming framework. In this paper, we propose a new Logic-Based Genetic Programming (LBGP) and show the effectiveness via computer experiments. In our research, we follow the ideas: (1) Each program as an individual consists of a set of claus...

• Type: Conference
• Year: 2004
• Where: GECCO
• Authors: Ken Taniguchi, Takao Terano
• Keywords: GECCO 2004, Genetic Programming, Logic-Based Genetic Programming, Redundant Clauses