一种改进的仿生算法研究与分析

本文对仿生算法中的神经网络和遗传算法进行了分析和总结,针对神经网络方法训练速度慢,识别效率低,以及遗传算法选优时的过早收敛的问题,结合神经网络方法和遗传算法各自的特性,提出了相应的改进算法。在不破坏单个神经元的输入权值的基础上,采用数据预处理的方法来减少输入层的个数,从而提高进化学习的能力。

An Improved Algorithm of Bionic Research and Analysis

In this paper the bionic algorithm of neural network and genetic algorithm were analyzed and summarized,aimed at slow speed of neural network training,recognition efficiency low,and genetic algorithm the optimum choice premature convergence problem,combined with neural network method and genetic algorithms of their respective characteristics,puts forward the improved algorithm.Without the destruction of single neurons based on input weights,adopt data pretreatment methods to reduce the number of input layers,so as to improve the ability of evolutionary learning.