Gene Expression Programming with Neutrality Used in Hardware Evolution

Aiming at the requirement of fast convergence in hardware evolution and the effect of evolutionary algorithm with neutrality on hardware evolution, this paper proposes a hardware evolution algorithm. Directed graphbased gene evolution programming, in which the gene expression and directed graph are integrated in hardware evolution and fitness distance correlation (fdc) is used to classify the difficulty of hardware evolution for the problems with or without neutrality. Two hardwareevolution problems are used to show the better performance of GGEP and effect of neutrality. The experiment results illustrate that GGEP has the highest success rate and average generations is less 4-20 times in second experiment than other algorithms. Convergence rate of algorithm with neutrality is faster than that without neutrality in first experiment and success rate is higher 20%-30% in second experiment.