基于模糊推理的进化规划用于神经网络设计

将模糊技术和进化计算技术融合,提出一种新的进化规划算法.该方法不仅能够加快普通进化规划算法的收敛性,而且能够有效地保证种群的多样性.将该方法用于神经网络的优化设计,计算机实验结果表明该方法性能稳定并减少了学习时间,较前人的研究成果更为有效.

Application of Evolutionay Programming Based on Fuzzy Inference in Neural Network

With adding the fuzzy technology into evolutionary computaion technology,proposed in this paper is a novel evolutionary programming .The method not only keep the population diversity but also has quiker convergence speed .It is applied to optimize the feedforaward neural network .The experimental results show that the new algorithm is stable ,it reduces the learning time and improve the generalization results in comparision with other well known methods.