Neural tree network model has been successfully applied to solving a variety of complex nonlinear problems. The
optimization of the neural tree model is divided into two steps in general: first structure optimization, and then parameter
optimization. One major problem in the evolution of structure without parameter information is noisy fitness evaluation, so
an improved breeder genetic programming algorithm is proposed to the synthesis of the optimization in neural tree network
model. Simulation results on two time series prediction problems show that the proposed optimization strategy is a potential
method with better performance and effectiveness.