Co-Evolving Hierarchical Programs using Genetic Programming

Abstract

The goal of this paper is to show that population size can be reduced by introducing different architectures (Automatic function Definition) in the initial population and the process of learning can be speeded up by having a separate sub-population for each automatic function. This provides the desired solutions of smaller sizes and also finds an effective way of scaling up the problem. A Standard Genetic programming. Genetic Programming with Automatic function defined functions (ADF) and Genetic Programming (with sub-population for each ADF) is implemented on 11-Multiplexer problem.