CAJ | 학술논문

结合神经网络方法,对入侵检测技术的聚类分析方法进行了研究和分析,探讨了在传统的对偶传播神经网络（Counter Propagation Networks,CPN）的基础上,引入基因表达式编程（Gene Expression Programming,GEP）对聚类进行优化,提出一种应用于入侵检测的CPN改进算法,该方法融合了GEP算法理念,克服传统的CPN网络算法中输入向量影响Kohonen层的连接权值,导致陷入网络抖动的缺点,提高了网络的收敛时间,得到了更好的网络聚类效果,并从实验中验证了算法在入侵数据分析中的有效性与优越性。
결합신경망락방법,대입침검측기술적취류분석방법진행료연구화분석,탐토료재전통적대우전파신경망락（Counter Propagation Networks,CPN）적기출상,인입기인표체식편정（Gene Expression Programming,GEP）대취류진행우화,제출일충응용우입침검측적CPN개진산법,해방법융합료GEP산법이념,극복전통적CPN망락산법중수입향량영향Kohonen층적련접권치,도치함입망락두동적결점,제고료망락적수렴시간,득도료경호적망락취류효과,병종실험중험증료산법재입침수거분석중적유효성여우월성。
Combined with the neural network method,an analysis of clustering in intrusion detection technology is conducted,and the optimization of the clustering by introducing the gene expression programming is discussed based on the counter propagation networks.And an improved CPN algorithm that is applied to the intrusion detection is proposed,which can overcome the problem of the input vector affecting the connection weight value in the Kohonen of the traditional CPN and in the mean time the networks will fall into the tremble.Furthermore,much better clustering result has been obtained,and the convergence time has been increased.Finally,the validity and superiority of the presented algorithm is demonstrated by experiment in intrusion data analysis.