神经网络算法在计算机网络安全中心中的应用研究  

Research on Application of Neural Network Algorithm in Computer Network Security Center

作　　者：胡兆文 HU Zhaowen(Qingdao Technical College,Qingdao Shandong 266555,China)

机构地区：[1]青岛职业技术学院,山东青岛266555

出　　处：《信息与电脑》2021年第16期77-80,共4页Information & Computer

摘　　要：为了对比BP神经网络代理模型的正确性和合理性,选取支持向量机SVM分别与多目标粒子群优化算法MOPSO和多目标遗传算法NSGA-II进行结合,并对目标函数进行评价优化。从实验数据可以发现,采用BP神经网络代理模型时得到的帕累托前沿较为均匀,而采用SVM代理模型时得到的帕累托前沿分布较为分散,甚至可能只出现某个极值点的情况,这个结果证明了BP神经网络代理模型对于多目标函数的优越性和准确性。选取ZDT、DTLZ测试函数进行实验,实验结果表明在仿真和耗时问题上BP神经网络代理模型结合进化算法具有一定的可行性。In order to compare the correctness and rationality of the BP neural network proxy model,this paper selects the support vector machine SVM to combine with the multi-objective particle swarm optimization algorithm MOPSO and the multiobjective genetic algorithm NSGA-II to evaluate and optimize the objective function.From the experimental data in this article,it can be found that when the BP neural network proxy model is used,the Pareto frontier obtained is relatively uniform,while when the SVM proxy model is used,the Pareto frontier obtained is more scattered,and there may even be only a certain extreme value.Point the situation.This result proves the superiority and accuracy of the BP neural network proxy model for multi-objective functions.In order to further verify this result,this chapter selects the ZDT and DTLZ test functions for experiments.The experimental data results show that the BP neural network is in the simulation and time-consuming issues.The feasibility of network agent model combined with evolutionary algorithm.

关 键 词：代理模型 BP神经网络 支持向量机SVM 遗传算法 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]