摘要
提出了一种带有影响因子的改进遗传算法并以此来优化前馈神经网络。染色体的每个基因都有一个影响因子,其不同取值体现了基因对整条染色体的不同影响程度。在遗传进化过程中,通过影响因子等遗传操作以达到对前馈神经网络的权值、阈值和结构优化的目的。仿真实验表明,该算法能够快速地确定神经网络的结构并且有效地提高了神经网络的收敛速度。
An improved genetic algorithm with influence factor was proposed to optimize feedforward neural network. There was an influence factor in each gene and different value of influence factor represents different influence to chromosome. In the process of genetic evolution, the goal of the optimization of weights, thresholds and network structure of feedforward neural network was achieved through genetic operation on influence factor etc. The simulation experiments show that this algorithm can determine the network structure of neural network quickly and improve convergence speed of neural network effectively.
出处
《计算机应用研究》
CSCD
北大核心
2007年第11期103-105,共3页
Application Research of Computers
基金
国防基础预研基金资助项目(S0500A001)
江苏省高校自然科学指导性计划项目(05JKD520050)
关键词
影响因子
改进遗传算法
前馈神经网络
优化设计
influence factor
improved genetic algorithm
feedforward neural network（FNN）
optimization
