Cartesian Ant Programming with node release mechanism

Abstract

Genetic Programming (GP) is one of the evolutionary algorithm that automatically creates a computer program. Cartesian GP (CGP) is one of the extensions of GP, which generates the graph structural programs. By using the graph structure, the solutions can be represented by more compact programs. Therefore, CGP is widely applied to the various problems. As a different approach from the evolutionary algorithm, there is the Ant Colony Optimization (ACO), which is an optimization method for combinatorial optimization problems based on the cooperative behavior of ants. By using pheromone communication, the promising solution space can be searched intensively. A number of ACO variants have been proposed for the various problem domains. One of them, ACO to automatic programming has been proposed recently. This new model, called Cartesian Ant Programming (CAP), is based graph representations in CGP with search mechanism of ACO. The connections of nodes are optimized by ant-based search instead of genetic operators. However, it is difficult to utilize the most part of given nodes as an effective node which are contained in the created program. In this paper, we propose a node release mechanism for CAP in order to utilize given nodes more efficiently. In the mechanism, specific nodes are set to unavailable at the start of the run. After certain step, unavailable nodes are released and all nodes become available. We compared the search performance of CAP with node release mechanism and normal CAP, and showed the effectiveness of our method.