基于基因表达式编程的递归函数挖掘

Mining Recursive Functions Based on Gene Expression Programming

传统基因表达式编程(GEP)无法发现递归函数。为此,分析了传统GEP算法在函数挖掘方面不足的深层次原因,提出了基于递归染色体的基因表达式编程算法GEP-RecurMiner和动态进化策略(DSCMS)。理论分析和实验证明了GEP-RecurMiner极大地扩充了传统GEP函数挖掘方法的求解空间,能精确地发现传统GEP无法发现的递归函数,同时实验表明动态进化策略有效地提高了GEP-RecurMiner函数挖掘算法的效率,挖掘成功率提高20%,平均进化代数下降10%。

Traditional Gene Expression Programming （GEP） is bare of discovering recursive functions. The limitation of function mining of the traditional GEP was analyzed. Revised algorithm GEP-RecurMiner based on recursive chromosomes and Dynamic Selection, Crossover and Mutation Strategy （DSCMS） based on best fitness were proposed. The theoretical proof and experiments showed that GEP-RecurMiner extremely extends the domain of function mining and can discover recursive functions. The experiments also showed that the performance of GEP-RecurMiner is improved by the combination of DSCMS. The number of average evolution generations decreases 10% , and the success rate increases 20%.