Using Genetic Algorithms to Evolve Behavior in Cellular Automata

Abstract. It is an unconventional computation approach to evolve solutions instead of calculating them. Although using evolutionary computation in computer science dates back to the 1960s, using an evolutionary approach to program other algorithms is not that well known. In this paper a genetic algorithm is used to evolve behavior in cellular automata. It shows how this approach works for different topologies and neighborhood shapes. Some different one dimensional neighborhood shapes are investigated with the genetic algorithm and yield surprisingly good results.

• Type: Conference
• Year: 2005
• Where: UC
• Authors: Thomas Bäck, Ron Breukelaar
• Keywords: Genetic Algorithm, Neighborhood Shapes, UC 2005, Unconventional Computation Approach