An improved representation for evolving programs

摘要: A representation has been developed that addresses some of the issues with other Genetic Program representations while maintaining their advantages. This combines easy reproduction linear inheritable characteristics tree by using fixed-length blocks genes representing single program statements. means each block will always map to same statement in parent and child unless it is mutated, irrespective changes surrounding blocks. method compared variable length gene used a clear improvement similarity between child. In addition, set list evaluation manipulation functions was evolved as an application new components. These have common feature they all need be 100% correct useful. Traditional Programming problems mainly optimization or approximation problems. The results are good but do highlight problem scalability more complex lead dramatic increase required evolution time.