基于跳跃基因的GA-BP算法研究  

GA-BP Algorithm Based on Jumping Gene

作　　者：生龙[1,2] 许林林 马晓雨 吴迪[1,2] SHENG Long;XU Lin-lin;MA Xiao-yu;WU Di(College of Information and Fleolriral F.ngineering,Heboi University of Engineering,Handan Hebei 0560381 China;Hebei Key Labratory of Security&Protection Information Sensing and Processing,Handan Hebei 056038,China;College of Civil Engineering,Hebei University of Engineering,Handan Hebei 056038,China)

机构地区：[1]河北工程大学信息与电气工程学院,河北邯郸056038 [2]河北省安防信息感知与处理重点实验室,河北邯郸056038 [3]河北工程大学土木工程学院,河北邯郸056038

出　　处：《计算机仿真》2020年第4期274-279,共6页Computer Simulation

基　　金：河北省自然科学基金(F2018402251);河北省高等学校科学技术重点研究基金项目(ZD2016017);河北省高等学校科学技术研究项目(ZD2018087);邯郸市科学技术研究与发展计划项目(1721203048)。

摘　　要：GA-BP学习算法往往会出现收敛速度慢,可能陷入局部极值的现象。针对以上问题,选取了自适应GA-BP(AGA-BP)算法,并在GA-BP算法和AGA-BP算法的基础上添加跳跃基因,称之为JG-GA-BP算法和JG-AGA-BP算法,用于解决分类问题。算法在遗传算法的基础上增加了跳跃基因算子,用于优化BP神经网络的结构参数,从而建立相应的神经网络拓扑模型。为验证添加跳跃基因后的学习算法的分类效果,将JG-AGA-BP算法、JG-GA-BP算法、AGA-BP算法和GA-BP算法的性能进行比较。以随机数、iris、wine、鲍鱼数据集的分类实验为例,研究结果显示出添加了跳跃基因的GA-BP算法的准确率和收敛速度都有一定程度的提高。The GA-BP learning algorithm tends to have a slow convergence rate and may fall into local extremes. Aiming at the above problems, we selected the adaptive GA-BP(AGA-BP) algorithm, and added the jumping gene based on GA-BP algorithm and AGA-BP algorithm, which is called JG-GA-BP algorithm and JG-AGA-BP algorithm and used to solve the classification problem. The jump gene based on GA was added in the algorithm to optimize the structural parameters of the BP neural network, and thus establishing a corresponding neural network topology model. In order to verify the classification effect of the learning algorithm after adding the jump gene, the performance of JG-AGA-BP algorithm, JG-GA-BP algorithm, AGA-BP algorithm and GA-BP algorithm were compared. Taking the classification experiments of random numbers, iris, wine and abalone data sets as an example, the results show that the accuracy and convergence speed of the GA-BP algorithm based on jump gene are improved to some extent.

关 键 词：跳跃基因 自适应算法 分类 神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]