Gene Expression Programming for Evolving Two-Dimensional Cellular Automata in a Distributed Environment.

This paper presents a novel distributed bio-inspired approach that uses Gene Expression Programming (GEP) to evolve transition rules for two-dimensional Cellular Automata (2D-CA). The 2D-CA are simulated in parallel using a master-slave distributed environment. The fitness function of the GEP ultimately measures the ability of a given CA to create a suitable solution for a complex Bioinformatics problem. To validate the proposed approach, extensive experiments were done dealing with a computationally expensive problem, that is considered to be one of the most important open challenges in Bioinformatics. Results of simulations show that the proposed approach was effective for the problem. Future works will investigate other distributed approaches of this approach, such as those based on General-Purpose Graphics Processing Units (GPGPU) or hardware-based accelerators. Finally, we believe that the method proposed in this work can be useful for other computational problems.