An Orthogonal Cartesian Genetic Programming Algorithm for Evolvable Hardware

摘要

Evolvable hardware (EHW) is facing the problems of scalability. Evolutionary algorithms often trap into local optima, or stalling in the later procedure. This paper analyses the difficulty of EHW. To improve the efficiency of Cartesian Genetic Programming (CGP), Neighborhood searching and orthogonal experiment design are tailed to an orthogonal mutation operator and a new Orthogonal Cartesian Genetic Programming algorithm is proposed. Demonstrated by experiments on the benchmark, the proposed Orthogonal Cartesian Genetic Programming can jump out of Local optima and decrease the stalling effect.