For keeping various of population dynamics in the process of evolution and improving the definition of solution of problems, a fine particle parallel algorithm which called CSGP(Cellular Symbol Genetic Programming) is proposed based on the SGP(Symbol Genetic Programming) algorithm and cellular automaton model. This algorithm has higher success rate and need less computation time. Experiments as symbolic regression and so on are given. The result indicates that the GSCP algorithm has more efficiency and precisely in cost and precision than GEPand SGP.%为维持进化过程中的种群多样性,并进一步提高求解问题的精确度,在SGP算法的基础上引入元胞自动机模型理论,提出一种能够实现具有细粒度并行的CSGP算法.该算法可提高求解问题的成功率以及减少进化代数,对比实验表明,CSGP算法较GEP算法和SGP算法在求解符号回归的问题上有较好的性能优势.
展开▼