CAJ | 학술논문

针对Kohonen神经网络模型网络入侵聚类正确率较低的问题，将入侵杂草优化(IWO)算法与Kohonen神经网络相结合，提出IWO-Kohonen聚类算法。利用IWO算法优化Kohonen神经网络的初始权值，训练Kohonen神经网络模型得到最优值。使用IWO算法增强算法的搜索能力，提高聚类正确率，并加快算法的收敛速度。实验结果表明，该算法与模糊聚类算法和广义神经网络聚类算法相比，分类正确率较高；与蚂蚁聚类算法和模糊C均值聚类算法相比，网络入侵检测率较高，误报率较低。
침대Kohonen신경망락모형망락입침취류정학솔교저적문제，장입침잡초우화(IWO)산법여Kohonen신경망락상결합，제출IWO-Kohonen취류산법。이용IWO산법우화Kohonen신경망락적초시권치，훈련Kohonen신경망락모형득도최우치。사용IWO산법증강산법적수색능력，제고취류정학솔，병가쾌산법적수렴속도。실험결과표명，해산법여모호취류산법화엄의신경망락취류산법상비，분류정학솔교고；여마의취류산법화모호C균치취류산법상비，망락입침검측솔교고，오보솔교저。
To improve the correct rate of the Kohonen neural network model for clustering of network intrusion, this paper combines the Invasive Weed Optimization(IWO) algorithm and the Kohonen neural network, and proposes IWO-Kohonen clustering algorithm. It uses IWO algorithm to optimize the initialized weights of the Kohonen neural network, and trains the Kohonen neural network model to calculate an optimal value. By using IWO algorithm, the search ability of the clustering algorithm is enhanced, which not only improves the correct rate of clustering, but also accelerates the convergence speed of the algorithm. Experimental results show that the proposed algorithm has higher correct rate comparing with fuzzy clustering algorithm and generalized neural network clustering algorithm, and it has higher detection rate and lower false alarm rate comparing with ant clustering algorithm and C-means clustering algorithm.