基于全连接循环神经网络的局域网异常流量检测技术  

Abnormal Traffic Detection Technology in Local Area Network Based on Full Connection Cyclic Neural Network

作　　者：郭小静[1] 李隘优[1] GUO Xiaojing;LI Aiyou(Minxi Vocational and Technical College,Longyan Fujian 364021,China)

机构地区：[1]闽西职业技术学院,福建龙岩364021

出　　处：《重庆科技学院学报（自然科学版）》2021年第5期107-112,共6页Journal of Chongqing University of Science and Technology：Natural Sciences Edition

摘　　要：针对局域网异常流量检测,提出了一种基于全连接循环神经网络的检测技术。全连接循环神经网络不仅考虑到神经元当前时刻的状态,也顾及到相邻时刻的神经元状态对模型数据训练产生的影响,确保神经元状态的连续性。为避免模型训练中梯度出现损失过快的情况,基于小批量梯度下降算法调整数据训练的收敛方向;当中间隐含层级神经元过多而增加系统消耗和模式复杂度时,利用GRU替代神经元保持神经网络的性能,并改善模型的训练和预测性能。实验结果显示,提出的检测技术的模型损失函数值与目标值趋近,且检测率指标值、假阳性率指标值和检测精度指标值都要优于现有检测技术。Aiming at the abnormal traffic detection of LAN,a detection technology based on full connection cyclic neural network is proposed.Fully-connected cyclic neural networks not only take into account the current state of neurons,but also the state of neurons at adjacent moments on model data training,so as to ensure the continuity of neuron states.In order to avoid the rapid gradient loss in model training,the convergence direction is adjusted based on the small-batch gradient descent algorithm.When too many intermediate hidden hierarchy neurons increase the system consumption and model complexity,GRU is used to replace neurons to maintain the performance of the neural network and improve the training and prediction performance of the model.The experimental results show that the model loss function value of the proposed detection technique is close to the target value;and the detection rate index,false positive rate index and detection accuracy index value are better than the existing detection techniques.

关 键 词：全连接 神经网络 局域网 流量检测 GRU 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]