摘要
基于BP神经网络的入侵检测方法因神经网络的初始网络运行参数是随机选择,存在容易陷入局部最优及收敛慢而导致检测准确率低的问题,提出一种基于CFA和BP神经网络的入侵检测方法 CFA-BPIDS.将BP神经网络的权值和阈值编码成CFA中的细胞个体,BP神经网络全局误差作为CFA的适应值,然后进行多次迭代,选择适应值最优的细胞个体作为BP神经网络的权值和阈值,最后将具有最优权值和阈值的BP神经网络应用在网络入侵检测中的检测模块.实验结果表明,该方法相比基于遗传算法和粒子群算法,优化BP神经网络的入侵检测方法提高了入侵检测准确率.
In the intrusion detection method based on BP neural network,because the initial network running parameters of neural network were random selection,it was easy to get into local optimal,slow convergence,and low detection accuracy.A method called CFA-BPIDS was proposed.It was based on CFA optimizing the BP neural network.The weight and threshold of BP neural network were encoded into the cell individuals in CFA.The deviation of BP neural network was used as the fitness value of CFA.Then the individual cell with the best fitness value was chosen as the initial weight and threshold of the BP neural network after iterations.Finally,the BP neural network model of CFA optimization was applied to intrusion detection.The simulation results showed that,compared with the BP neural network model which was optimized by genetic algorithm and particle swarm optimization,the proposed method improved detection accuracy.
作者
凌捷
黄盛
LING Jie;HUANG Sheng(School of Computer,Guangdong University of Technology,Guangzhou 510006,China)
出处
《郑州大学学报（理学版）》
CAS
北大核心
2018年第3期1-6,共6页
Journal of Zhengzhou University:Natural Science Edition
基金
广东省科技项目(2014B090901053
2015B010128014
2015B090906015
2016B010107002)
广州市科技计划项目(201604010077
201604010048)
关键词
CFA
BP神经网络
入侵检测
遗传算法
粒子群算法
cuttlefish algorithm
BP neural network
intrusion detection
genetic algorithm
particle swarm optimization
