基于强化LSTM的网络安全态势预测方法

Network Security Situation Prediction Method Based on Sigmoid Weighted Strengthened LSTM

网络安全态势预测作为网络安全态势感知的重要组成部分,描述的是安全态势随时间动态变化的行为,具体是根据历史态势值预测未来态势值。为了提高网络安全态势预测准确率,提出一种基于sigmoid加权强化机制的长短期记忆的网络安全态势预测模型。该方法首先利用LSTM神经网络挖掘网络安全态势数据之间的时间相关性;引入sigmoid加权线性单元来处理反向传播中的梯度问题,将输入值与sigmoid激活函数相乘,从而强化LSTM神经网络结构,提高预测的精度。然后采用布谷鸟搜索算法进行超参数寻优,提高网络训练时间。最后利用国家互联网应急中心的网络安全态势数据对该模型进行验证,仿真实验结果验证了该方法的合理性,以及该模型具有更快的收敛速度和更小的误差,提高了网络安全预测能力。

As an important part of network security situation perception,network security situation prediction describes the behavior of dynamic changes of security situation over time,and specifically predicts the future situation value according to the historical situation value.In order to improve the accuracy of network security situation prediction,a network security situation prediction model based on sigmoid weighted strengthening mechanism of long and short memory is proposed.This method firstly uses LSTM neural network to mine the temporal correlation between network security situation data.Sigmoid weighted linear unit is introduced to deal with the gradient problem in back propagation,and the input value is multiplied by sigmoid activation function,so as to strengthen the structure of LSTM neural network and improve the prediction accuracy.Then cuckoo search algorithm is used to optimize the super parameters to improve the network training time.Finally,the network security situation data of The National Internet Emergency Response Center is used to verify the model.The simulation has verified the rationality of the proposed method,and the model has a faster convergence speed and a smaller error,which improves the network security prediction ability.